1599
Anal. Chem. 1983, 55, 1599-1603
6
composition of the sample in the detector or parts leading to the detector. The minimum detectable quantity is obviously related to the electron capture coefficient. Consequently, it will also be a function of the capture mechanism and the temperature. For 1,3,5-trichlorobenzene, and the highest temperature, picogram quantities can be detected. Registry No. 1,3,5-Trichlorobenzene,108-70-3;benzonitrile, 100-47-0;naphthalene, 91-20-3;benzaldehyde, 100-52-7;benzyl acetate, 140-11-4; 3-nitrotoluene, 99-08-1; 1,3-dichloropropane, 142-28-9; 2-chlorotoluene, 95-49-8; chlorobenzene, 108-90-7; bromobenzene, 108-86-1; 1,2-dichlorobenzene, 95-50-1; iodobenzene, 591-50-4; m-dibromobenzene, 108-36-1; 1,1,2,2-tetrachloroethane, 79-34-5; nitrotoluene, 1321-12-6; nitrobenzene, 98-95-3;p-chloronitrobenzene,100-00-5.
i i , , ! - 3 - 2 - 1
0
1
2
3
4
Log w
[.gl Figure 4. Constant currisnt response vs. amount injected at various temperatures (K)for benzaldehyde (A)and 1,3,54richIorobenzene (0).
detectable quantity for a dissociative compound, 1,3,5-trichlorobenzene, and a nondissociative compound, benzaldehyde. The responses are plotted as relative area vs. the amount injected on a log-log plot at different temperatures in Figure 4. The analytical parameters obtainled from the data are tabulated in Table 11. The maximum linear dynamic range is 3 orders of magnitude. For benzaldehyde, there is no linear range at the highest temperature. This could be due to de-
LITERATURE CITED Lovelock, J. E.; Llpsky, S. R. J . Am. Chem. SOC. 1980, 82, 431-433. Lovelock, J. E. Anal. Chem. 1981, 33, 162-177. Wentworth, W. E.; Chen, E. C. M.; Lovelock, J. E. J . Phys. Chem. 1988, 70, 445-458. Fenimore, D. C.; Davis, C. M. J . Chromatogr. Scl. 1970, 8 , 519-523. Maggs, R. J.; Joynes, P. L.; Davles, A. J.; Lovelock, J. E. Anal. Chem. 1971, 43, 1966-1971. Patterson, P. L. J . Chrofytogr. 1977, 134, 25-37. Grimsrud, E. P. I n Electron Capture-Theory and Practice Chromatography”; Zlatkis, A., Pooie, C. F., Eds.; Elsevier: New York, 1981; pp 91-117. Wentworth, W. E.; Chen, E. C. M. J . Chromatogr. 1979, 786, 92-1 18. Wentworth, W. E.; Chen, E. C. M. I n “Electron Capture-Theory and Practice in Chromatography”; Zlatkis, A., Poole, C. F., Eds.; Elsevier: New York, 1981; pp 27-68.
RECEIVED for review February 16,1983. Accepted May 2,1983.
Quantitative Analysis without Analyte Identification by Refractive Index Detection Robert E. Synovec and Edward S. Yeung” Department of Chemisfty and Ames Laboratory, Iowa State University, Ames, Iowa 500 1 I
Contrary to the accepted notion that qualitative analysis must precede quantitative (analysis,we show that one can determlne the volume fraction of an arbitrary anaiyte In a flowing s identity. The physlcai and chemsystem without knowing H ical properties of the anaiyte are thus unknown, and no working curve can be established. The scheme Is based on the refractive index detector, whlch produces two distinct responses for the same analyte when two separate eluents are used. The informiation obtained can be used to predlct the volume fractlon of the analyte, as well as Its refractive index. Experimental verlflcation of thlts scheme Is reported.
Chemical analysis deals with the solutions to scientific questions through the identification ,and the quantitative determination of the composition of matter. It is generally accepted that the former precedes the latter. That is, one must identify or specify the species of interest before one can determine its concentration. This is because all analytical methods are based on some particular physical and/or chemical property of the species. The experimentalobservable must be calibrated against this particular property so that a concentration can be deduced. Analytical working curves thus
serve to provide the needed calibration, but they can only be constructed if the identity of the species is not in doubt. There are certain situations where it is desirable to know the concentrations of the components before any attempts at identification. One example is the assay of supposedly pure material. There, one tries to determine the type and the amount of each impurity present, the latter being of primary concern. Since quantitative methods are generally species specific, it is difficult to be sure that all possible impurities have been searched for. If however one can first ascertain the amount of all impurities present, the scope of the problem becomes much more tractable. Another example is the control of pollution emission and waste discharge. Knowing the concentrations of any foreign matter released without first requiring speciation is advantageous. A third example is forensic chemistry, where finding out whether any contaminant exists at all is an important fist step. A fourth example is in organic synthesis, where it may be desirable to know the yields of the various reaction products on the microscale, even when these products cannot be identified. And, when the products are identified, it may not be possible to isolate sufficient quantities of each to use traditional analytical calibration curves. It is therefore appropriate to pose the question whether qualitative analysis is a prerequisite for quantitative analysis.
0003-%700/83/0355-1599$0’1.50/0 0 1983 American Chemical Society
1600
e
ANALYTICAL CHEMISTRY, VOL. 55, NO.
9,AUGUST 1983
By quantitative analysis we mean the determination of concentrations in terms of volume fraction, weight fraction, or mole fraction of a species and not secondary properties like oxygen uptake or total carbon. Naturally, the first two can always be determined in conjunction with some separation procedure, as long as the quantities are large enough for direct measurements, for example, in preparative scale LC. We shall restrict our discussion to trace analysis and microanalysis. There are a few known analytical methods that come close to providing quantitative information for an unidentified analyte. Coulometry relates electrochemical equivalents to an observable that can be independently calibrated-the current. To obtain the concentration, however, one needs to know the change in the number of electrons in the reaction, as well as the current efficiency of the system. Mass spectrometry can in principle determine the number and the molecular weight of a species simultaneously, but the choice of conditions affects the ionization efficiency and the extent of fragmentation, and thus the quantitation. Furthermore, the same number of ions of different types need not produce the same ion count in the detector. The electron capture detector for gas chromatography (GC) has been suggested as an absolute gas-phase coulometer ( I , 21, but positive ions, dissociation, and instrumental effects create problems. Besides, only compounds with very high capture cross sections are suitable for applying this scheme. The scheme we propose can be explained qualitatively by using the following example. Suppose a refractive index (RI) detector is used to monitor the elution of an analyte in a flowing system. Let us further suppose that no signal (from base line) is observed even at the highest sensitivity setting of the detector. On changing to a different eluent, however, a detector response is obtained for the same injected amount. The lack of response in the first experiment is in fact an important piece of information-that the RI of the analyte is essentially the same as that of the first eluent. If the RI of both eluents are known from independent measurements, this then allows us to calculate the concentration of the analyte. In general, it is fortuitous if in fact no response is obtained from a particular eluent. But, as long as the two eluents have different RI’s, it can be seen that the two experiments provide two equations with two unknowns, the RI and the concentration of the analyte, and the latter can be uniquely determined.
THEORY It is necessary to start from first principles to see how RI depends on the components in a mixture. From the basic relations among the electric field, the electric displacement vector, and the polarization, one obtains the Clausius-Mosotti relation (3)
where x is the susceptibility per molecule, M is the molecular weight, No is Avogadro’s number, p is the density of the material, and E’ is the dielectric constant. It can be shown that, except at high densities or high field intensities, neither of which is true in LC, the susceptibility of a mixture is additive (3)
x
=
Cxixi
(2)
1
where x i is the mole fraction of component i with susceptibility xi. It may be noted that for sufficiently low field strengths or high temperatures
D? X L = ai + 3kT
_ I
(3)
where ai and Di are the polarizability and the permanent dipole moment of the species i, k is the Boltzmann constant, and T i s the absolute temperature. Now, the volume fraction of the component i, Ci, is given by
(4) The refractive index of component i, ni, is given by ni =
a
(5)
where pi is the permeability of the medium. But since p’ is very nearly unity (typically deviating by less than one can replace tri with n: in eq 1. Grouping eq 1 , 2 , 4 , and 5, one has
with n being the RI of the mixture. Now consider the case of a binary mixture, which is composed of an analyte species and an eluent. A t any particular instant, the measured RI response is determined by the volume fraction of the analyte of RI n, in the flow cell, C,, and the volume fraction of the eluent in the flow cell, (1- C,). Using eq 6 and the subscript 1 for the eluent, one obtains
n2- 1
n2 + 2
n,* - 1 -
- I _ _ _ -
+
n12 2
.,[
-]+
nX2- 1 - n12- 1 -
+
nZ2 2
n12
2
(7)
Combining the two terms on the left of eq 7 , one gets 3(n2n,2)/[(nz + 2)(n,2-t 211, which in turn equals 3Anl(n+ n l ) / [ ( n 2 2)(n? 2)]. Anl is in fact the experimentalobservable from any of the differential RI detectors. Note that for low concentrations, errors much less than 0.1 % are expected if all n’s are now replaced by nl’s. Rearranging and redefining terms
+
+
AnlKlr = C J F , - F,)
(8)
+
where Fi (n? - l ) / ( n i 2 2 ) , and K[ 6n,/(n? + 2)2. So, eq 8 shows that by using two eluents 1 and 2 for the same injected quantity of the analyte AnlKl‘ - An2K2’
c, =
F2 - Fl
and the concentration is thus determined. To relate these to the volume fraction of the analyte at injection, one must integrate these C, values over the entire detector response curve, which is shaped like a peak, and then multiply by the ratio of the total peak volume to the injection volume. It is therefore much more convenient to think in terms of peak areas, in units of RI-volume. This can be derived from the RI vs. time detector response curve that is normally obtained by measuring the eluent flow rate. After this, one can insert eq 9 into eq 8 and determine the value n,. It is important to note that the denominator in eq 9 is the order of (n2- nl) and that the K’values are the order of unity. So, no loss in sensitivity or in significance is expected in applying this scheme. In fact, the choice of two eluents of substantially different RI’s guarantees that a signal will be obtained in at least one of the eluents for any analyte. Also, it is important to use the correct signs for Ani throughout, otherwise an analyk with n, intermediate between nl and n2will incorrectly show a cancellation in the numerator of eq 9. One can expand further on this concept. It is known that RI depends on the temperature, the pressure, and the wavelength of light used. It is also known that commercial detectors for RI show a response that is dependent upon the RI of the eluent ( 4 ) ,unlike RI detectors based on interfero-
ANALYTICAL CHEMISTRY, VOL. 55, NO. 9, AUGUST 1983
metry (5,6). It may also be necessary to use different eluent flow rates for the two experiments. It 1i3 therefore desirable to rewrite eq 8 in termri of the observed !signal,S, which is an area in arbitrary units SlKl = V,(F, - F J (10)
13878 /
second eluent, a different conversion factor for response, and the flow rate of the second eluent. From eq 10 and 11
We can now obtain two other detector response curves, injecting equal amounts (for example, equal volumes with equal volume fractions) V of each eluent into the other under the same conditions. So
S3K1 = V(F2- F1)
(1.3)
SAKz = V(F1- F2)
(14)
and From these, we can see that
H:,/KZ = -S,/S,
K
v, = v(
2 + 2)
(1.6)
In other words, the volume fraction of the analyte injected is determined without any knowledge of the instrumental response factor or the identities of the two eluents and the analyte! The only requirement is that the same conditions are used throughout for these measurements. Again, no loss of sensitivity or significance is expected. To then solve for F, and thus n,, it is necessary to know the two values nl and n2 From eq 10 and 11, one has
F, - F1 .-SiK1 = SZK2
F, - F2
(17)
So, making use of eq 15, one can solve for F,.
EXPERIMENTAL SECTION All reagents and eluents used are reagent grade materials without further purification. A conventional chromatographic system was used, although not necessary ,for the demonstration of the concept. It consisted of a reciprocating pump (Milton Roy, Riviera Beach, FL, Model 196-0066),a 25 cm X 4.6 mm, 10-pm CI8 column (Alltech, Deerfield, IL), a 5-pL sample loop at a conventional injection valve (Rheodyne, Berkeley, CA, Model 7010), and a commercialRI detector (Waters Associates, Milford, MA, Model R401) with the reference cell used in the static mode filled with the eluent being used. A flow rate of 0.60 mL/min is used throughout. Solutionswith specified volume fractions are made by pipetting a well-defined volume of'the minor component into a volumetric flask and then fiiing to the mark with the major component. The output of the RI detector (10 mV full scale) is connected to a digital voltmeter (Keithley, Cleveland, OH, Model 160B), the analog output of which is in turn connected to a computer (Digital Equipment, Maynard, MA, Model PDP 11/10 with
I
1.3839
/
\
\ \
'1.4952
I
Flgure 1. R I dependence on volume fraction for benzene in heptane: solid horizontal line, linear interpolation; solid curve, true dependence; dashed lines, limiting slopes.
LPS-11laboratory interface). The computer takes readings every 0.05 s and averages each set of 10 before storing the information. Typically about 100 of these averaged data points define an analyte peak. The area is determined by locating and determining the peak maximum and then summing all points that are more than 1.0% of this peak value above the base line. The numerical values of these areas are then used directly as Sidefined earlier. All areas are determined by using multiple injections (three or more) and are found to be reproducible to rt2.5% (relative standard deviation). The linearity of the attenuation settings on the RI detector was determined by injections of successively diluted samples and was found to be &2.5%from 32X to 2X, which covers all of the scales used in this work. Measurementsof RI for various mixtures were performed with a standard Abbe refractometer (Spencer Optical, Buffalo, NY) after calibration with pure solvents.
(15)
Combining eq 12, 13, and 15, the result is
\ ' \ I
I
where K is a new constant grouping together the old K', the particular flow rate for this eluent, the injection volume, and the conversion factor from Anl to the unnts of the signal (such as a voltage or a number in the computer), and V , is the volume fraction of the anal@ at injection. As long as the same light source, the same temperature, and the same pressure are used throughout, F, and Fl are true conutants. For the same injection using a second eluent
K 2 is a different constant because it includes the RI of the
1601
RESULTS AND DISCUSSION To appreciate the implications of eq 7 and 8, we have plotted the actual dependence of RI of a mixture of heptane and benzene as a function of the volume fraction of benzene in Figure 1. The ordinates are shifted on the left compared to the right to facilitate visualization. The solid horizontal line is a linear interpolation of the RI's for the two pure solvents and is commonly assumed in elementary discussions of applications of refractometry (7). The curved line is the actual dependence from eq 7. It can be seen that the linear interpolation does extremely poorly throughout. Note that the correct curvature shown is not due to a nonideal solution. A nonideal solution simply changes the implication of the volume fraction C, and in a sense makes the horizontal scale nonlinear. The deviation from an ideal solution will introduce a corresponding amount of error in our determination. Unless either the analyte or eluent is extremely polar, one can safely assume that an ideal solution is formed. In liquid chromatography (LC), for instance, this assumption is made implicitly in all measurements. Otherwise, the concentration a t the detector cannot be related to the concentration of the injected sample in a linear fashion. The dashed lines in Figure 1 are the tangents to the curve for small concentrations of each of the components. From eq 8, the two slopes are given by (F, - Fl)/K,' and (Fl - F,)/KJ, respectively. The fact that these slopes are different for each of the two solvents used allows us to determine the value C, independent of the particular value of F,. Using a standard refractometer, we have independently confirmed linearity and the limiting slope, as specified by eq 8 for cases with the minor component present below 4 % by volume for a number of binary mixtures involving combinations of hexane, benzene, CHCl,, CC14,and
cs2.
The raw data, averaged for each set of multiple injections, are shown in Table I. Full scale corresponds to 10 mV from the detector module, amplified to 1.0 V by the digital volt-
1602
ANALYTICAL CHEMISTRY, VOL. 55, NO. 9, AUGUST 1983
Table I. Areas of Eluted Peaks injected volume eluent analyte fraction heptane heptane heptane heptane heptane heptane heptane heptane heptane benzene benzene benzene benzene benzene benzene benzene benzene benzene
benzene benzene benzene benzene benzene CCl, CCl, CHCl, CHCl, heptane heptane heptane heptane heptane CCl, CC1, CHCl, CHCl,
atten- area uation (~10,)
0.02 0.01 0.005 0.00167 0.00167 0.03 0.003 0.03 0.003 0.02 0.01 0.005 0.00167 0.00167 0.03 0.003 0.03 0.003
32 16 8 4 2 32 4 16 2 32 16 8 4 2 32 4 32 4
3.86 3.92 3.88 2.65 5.25 4.12 3.13 5.54 4.59 -4.75 -4.76 -4.70 -3.08 -6.14 -2.02 -1.66 -3.39 -2.76
Table 11. Predictions for Simulated Mixtures sam- true true ple C, RI CC1, 0.030 1.4601 0.0030 1.4601 CHCI, 0.030 1.4459 0.0030 1.4459
eq 9,8 cx RI 0.0298 1.4680 0.00286 1.4661 0.0284 1.4461 0.00296 1.4436
eq 16,17,15 c, RI 0.0299 1.4679 0.00287 1.4647 0.0286 1.4435 0.00295 1.4425
meter, and converted to a numerical value of 2048 by the A/D converter in the laboratory interface. The 10 measurements in Table I involving heptane and benzene only can be used to calibrate the attenuation settings of the refractometer and to determine the goodness of eq 8. In order, the five values involving benzene injected into heptane give calibrations for full scale at each attenuation of 11.5,5.66,2.86,1.40,and 0.71, in RI units. The five values involving heptane injected into benzene give calibrations for full scale at the corresponding attenuation of 10.2,5.08,2.57,1.31, and 0.66,in 10” RI units. Linearity of the scales and of eq 8 is thus confirmed to f2.5%. The ratios of the two sets of RI scales are consistent with a slight nonlinearity in the particular instrument design. With these calibrated scales for RI, we can now relate the areas for CCll and for CHC&to RI-volume units. Since the injected volume is 0.005 mL and the flow rate used provides also 0.005 mL per integration interval, the areas in Table I are simply multiplied by the full scale equivalent RI and then divided by the range of the AID converter, 2048. We find from published work (8) that nheptane = 1.3878and nbenzene = 1.5011. This gives Fl = 0.2359,F2= 0.2947,K,’ = 0.5402,and K,’ = 0.4979. So,the concentrations of the “unknowns” can be found by using eq 9 and are tabulated in Table 11. Compared to the “true“ volume fractions, the agreement is good, with an average relative error of 3%, which is consistent with the standard deviations of the individual area measurements in each set of multiple injections. Also tabulated in Table I1 are the predicted values of the RI of the unknowns using eq 8 and the RI of benzene. The agreement with known values is again good, especially since the wavelength used in the RI detector is not the sodium D line and the temperature of the eluent is not controlled to be at 20 “C. Further, we note that, on careful inspection, the sample of CHC13 contains an unknown impurity. This impurity was not resolved chromatographically by the system used. This is probably a preservative for the reagent. So, even though the volume fraction is correctly predicted, a slight systematic error in the RI calculated is expected. By use of eq 8 and the RI of n-heptane,
similar values are obtained for the unknown RI’s. However, it is known that the eluent we used actually contained about 2% of impurities, so that any comparison is inappropriate. The problems with the above procedure are now obvious. It is necessary to calibrate the instrument in actual RI units (which typically changes from one eluent to the next) and to maintain conditions so that the temperature, pressure, and wavelength all correspond to the chosen values for nl and n2. If this quantitation concept is applied to LC, this concern becomes even more serious if a mixed solvent provides the best chromatographiccondition (9),and an accurate value for n is not available. All these problems can be avoided if instead eq 16 is used. We have calculated in this manner the concentrations for the “unknowns”. These are shown in Table 11. The agreements with the true values are again within the relative standard deviations of each set of area measurements. In fact, the average relative error of 2.7% is slightly better than that obtained by using the previous procedure, as expected. The calculated values for the n’s for the unknowns are found by using eq 15 and 17 and are listed in Table 11. Again, the comparison with the true values is good in view of the discussions in the preceding paragraph. The application of this quantitation concept for LC should be considered. For instance, it is possible to extend this concept to the case of a mixture of analytes x and y, in which the analytes are not chromatographically resolved. One can obtain the analogue to eq 8 AnlKl’ = C,F, CS;;, - (C, C J F , (18)
+
+
It can be seen that by following the same procedure described above, one can uniquely determine the total concentration, C, + Cy, contributing to the particular chromatographicpeak, regardless of the individual refractive indexes. In this case, the calculated refractive index will be intermediate between those of components x and y and will not have any significance because of its concentration dependence. We note that if one uses four eluents (different nJ to elute this sample, there will then be available four equations of the form in eq 18. However, these equations are not independent and can be reduced to only two independent equations. So, the four unknowns, C,, Cy,F,, and Fy, cannot be solved for uniquely. The calculations above require that the chromatographic peaks must be correlated between the chromatograms obtained with the two eluents, i.e., the order of elution must be known. In gel permeation chromatography using selected pairs of solvents, this is not a major problem. In other cases, one can obtain a third chromatogram with an eluent with a RI different from the other two. There are a total of N! possible different elution orders for a sample with N components in the first two chromatograms. If one now uses the above procedure to determine V, and n, based on the first pair of chromatograms, one can predict the response, S, in the third chromatogram for every one of the N peaks. If the elution order changed in any of the three eluents, consistency will not be achieved. It is possible for the computer to go through all N! combinations until consistency is obtained in the predictions. This way, the elution order for each of the N peaks in each of the three eluents can be uniquely determined. It is easy to see that the only time this consistency test fails is if both the concentration and the RI for two or more of the components are identical. In this unlikely case, one does not worry about the individual concentrations anyway. Applications of this kind of consistency test include the optimization of chromatographic conditions in LC, when three chromatograms rather than 3(N - 1) chromatograms are needed to check the elution order. An intriguing question is whether there are other similar methods in chemical analysis that can be adapted to this scheme. An inspection of the basic operations shows that the
Anal. Chem. 1983, 55, 1603-1605
experimental observable must be related (linearly, if possible) simultaneously to some property of the analyte and the environment (the eluent). It must be possible to change the environment while keeping the concentiration of the analyte and the property of thLe analyte the same. For this reason, some chromatographic method that allows a change of eluent is the simplest solution. Following the procedure here, the peak area is meaningful regardless of the change in actual separation conditions (different k's or different number of theoretical plates), as long as one can correlate the peaks in each chromatogram. The detector must be a true differential detector, so that the contributions of the pure environment can be used as the base line. Otherwise, one will end up subtracting two large numbers to detect the small change in the experimental observable, losing sensitivity and significance at the same time. The other common differential detector in LC is the absorption detector. It satisfies all of the conditions outlined above. To be adapted to this scheme, one must use at least two eluents with different molar absorptivities at the wavelength of interest (e.g., benzene and hexane at 254 nm). The difficulty is that when the highly absorbing eluent is used, so little light reaches the sensor of the detector that the normal electronics will not function. So, some redesign of the commercial UV absorption detector for LC must first be made. Then, it can be seen that mole fraction (rather than volume fraction) becomes the appropriate unlit for concentration. Technical problems not withstanding, the micropolarimeter for LC (IO) can be used in conjunction with optically active eluents, e.g., (f)3-mt~thylcyclohexanoneand a nonchiral analogue. The differential measurement is essentially accomplished by setting the appropriate mechanical null for the analyzer. There, weiglht fraction is the relevant unit. In gas chromatography, the two detection methods that fit this scheme are the gas density balance (11) and the thermal conductivity detector (12). The former is based on a very complicated and delicate design and has relatively large volumes and poor sensitivity. So, even though the weight and the molecular weight can both be determined by using our scheme, it is not likely to be of great value. The latter involves very well-developed instrumentation, but the dependence on
1603
molecular properties is very complicated. For low concentrations and molecular weights very different from that of the eluent, however, some approximations can be made to fit our scheme. Mole fraction is again the appropriate unit of concentration. The universal nature of its response is once again attractive. Work toward demonstrating some of these other concepts is in progress in our laboratory. Finally, it should be particularly interesting to apply this scheme to gel. permeation chromatography, where many eluents produce the same chromatogram. The molecular size information in the chromatogram can be used in conjunction with the volume fraction determined here to give a count of the molecules. This concept should work even for poorly resolved chromatograms (fairly continuous distribution of molecular sizes), by using eq 16 for each well-defined "slice" of the chromatogram. In the characterization of polymers or of fossil fuels, this is valuable new information. All these can be performed by using analytical scale LC, with instrumentation that is available in most analytical laboratories.
LITERATURE CITED (1) Lovelock, J. E.; MagQS, R. J.; Adlard, E. R. Anal. Chem. 1971, 43, 1962- 1965. (2) Lovelock, J. E. J. Chromatogr. 1974, 99,3-12. (3) Hirschfelder, J. 0.; Curtlss, C. F.; Bird, R. 8. "Molecular Theory of Gases and Liquids"; Wiley: New York, 1964; pp 852-871. (4) Yeung, E. S. I n "Advances in Chromatography"; Giddlngs, J. C., Grushka, E., Brown, P. R., Eds.; Marcel Dekker: New York, 1983. (5) Woodruff, 5. D.; Yeung, E. S. Anal. Chem. 1982, 54, 1174-1178. (6) Woodruff, S. D.;Yeung, E. S. Anal. Chem. 1982, 54, 2124-2125. (7) Skoog, D. b . ; West, D. M. "Principles of Instrumental Analysis", 2nd ed.;W. A. Saunders: Phlladelphia, PA, 1980; p 374. (8) "CRC Handbook of Chemistry and Physics"; Weast, R. C., Astle, M. J., Eds.; CRC Press: Boca Raton, FL, 1978. (9) Snyder, L. R. J. Chromatogr. Sci. 1978, 16, 223-234. (IO) Yeung, E. S.; Steenhoek, L. E.; Woodruff, S. D.; Kuo, J. C. Anal. Chem. 1980, 52, 1399-1402. (11) Martin, A. J. P.; James, A. T. Blochem. J. 1958, 63, 138-143. (12) Llttlewood, A. B. Nature (London) 1959, 164, 1631-1632.
RECEIVED for review December 3,1982. Accepted May 9,1983. R.E.S. thanks the Dow Chemical Co. for a research fellowship. The Ames Laboratory is operated for the U S . Department of Energy by Iowa State University under Contract No. W-7405-eng-82. This work was supported by the Office of Basic Energy Sciences.
Determinatiion of Selenium in Marine Sediments by Gas Chromatogiraphy with Electron Capture Detection K. W. Michael Slu" and Shier S. Borman
'
Division of Chemistry, National Research Council of Canada, Ottawa, Ontario, Canada, K7A OR9
Selenium concentrations in marlne sediments were determined by gas chromatography with electron capture detection. The sedlments were dissolved in an acid mixture. Selenium was converted lnto ~i-nitrop~azrse~eno~, extracted Into toluene, and introduced Into the chromatograph. The detection limit was 0.2 pg alf Se Injected or 20 ng of Se/g of sediment. The standard deviation was about 7%.
1,2-Diaminobenzene(0-phenylenediamine) and its derivatives react selectively and quantitatively with selenium(1V) lNRCC 21290.
to form piazselenols that are both volatile and stable. Trace levels of selenium in materials as diverse as plant and animal tissues, metals, and natural waters have been determined success full^ as piazselenols by means of gas chromatography with electron capture detection. These analyses have been reviewed by T6ei et al. (2). No work on sediments or geological materials, however, has been reported in the literature. This paper describes the determination of selenium as 5-nitropiazselenol in marine sediments.
EXPERIMENTAL SECTION A Varian Aerograph Model 1200 gas 63Ni electron capture chromatographequipped with a T~~~~~ detector (ECD)was used. The column was a 2-m borosilicate tube packed with 3% OV-225 on Chrornosorb W, SO/lOO mesh. It was Instrumentation.
0003-2700/83/0355-1603$01.50/0 Published 1983 by the American Chemical Soclety
