870
ANALYTICAL CHEMISTRY, VOL. 51,
NO. 7,
JUNE 1979
Table 11. Removal of Halide Interference with Derivatization of Ambient Nitrate Samples b y Means of Silver Acetate Addition PdmL un- by colorknown imetry 03-5 03-6
1.88
03-7
3.76
03-8 04-1 04-2
2.29 0.44 0.45
04-3
0.60
1.89
% b y GC-ECD
-
n o silver
acetate
nitration method silver acetate addeda 82
43 74 63 ND 79 81 ND
89 93 93
-___ x = 68 i 16%
111 ND 70
y =90 1
14%
a An amount of AgAc was added equivalent to the amount of nitrate found by the colorimetric method.
to ambient nitric acid determination is also in progress.
LITERATURE CITED (1) D. H. Fine, D. P. Rounbehier, and P. E. Oettinger, Anal. Chim. Acta. 78. 383 (1975).
(2) C. W. Spicer. Atmos. Environ., 11, 1089 (1977). (3) J. B. Muilin and J. P. Riley, Anal. Chim. Acta, 12, 464 (1955). (4) J. Mulik, R. Puckett, D. Williams, and E. Sawicki, Anal. Left.. 9 , 653 (1976). (5) E. Sawicki, H. Johnson, and T. W. Stanley, Anal. Chem., 35, 1934 (1963). (6) A. Oien and A. R. Selmer-Olsen, Analyst (London), 94, 888 (1969). ( 7 ) G . Norwitz and H. Gordon, Anal. Chim. Acta, 89, 177 (1977). (8) W. D. Ross, G. W. Butler, T. G. Duffy, W. R. Reyh, M. T. Wininger, and R. E. Sievers, J . Chromatogr., 112, 719 (1975). (9) J. W. Tesch, W. R. Rehg, and R. E. Sievers, J . Chromatogr.. 126, 743 (1976). (10) S. A . Hubbard, E. Wlttgenstein, and E. Sawicki, Paper ENVR-9, 172nd National Meeting, American Chemical Society, San Francisco, Calif., Aug. 29-Sept. 3, 1976. ( 11 ) R. J. Hare, M. T. Winninger, W. D. Ross, J. Tesch, and R. E. Sievers, Paper ENVR-99, 176th National Meeting, American Chemical Society, Miami Beach, Fla., Sept. 10-15, 1978. (12) C. A. Clemons and A. P. Altshuller, Anal. Chem., 38, 133 (1966). (13) N. K . McCallum and R. J. Armstrong, J . Chromatogr., 78, 303 (1973). (14) G. M. Le Fave, J . A m . Chem. Soc., 71, 4148 (1949). (15) C. L. Cwn, W. G. Biucher. and M. E. Hill, J . Org. Chem., 38, 4243 (1973).
RECEIVEDfor review October 13, 1978. Accepted February 9, 1979. The authors gratefully acknowledge the support of this work by the Division of Chemical Sciences, United States Department of Energy under contract No. EY 76-C-02-0016.
Characterization of Isomeric Compounds by Gas and Plasma Chromatography D. F. Hagen 3 M Central Research Laboratories, St. Paul, Minnesota
55 10 1
The combination gas and plasma chromatographic system has demonstrated the capability of dlfferentiatlng between isomeric compounds. Reduced ion mobility data are inversely proportional to average molecular collisional cross sections. Meta substiiuted positional isomers are generally larger than the para and ortho isomers. 4,4’-Dipyridyl has a cross sectional value approximately 8 A* larger than the 2,2’-dipyridyl species. Geometric cis-trans isomers also give large differences in cross sections as calculated from ion mobility data. Although errors may exist in the absolute value of a cross section, differences as small as 0.2 A* in a given isomeric series are measurable with the system and assumptions described.
T h e plasma chromatograph (Franklin GNO Corp., West Palm Beach, Fla.) has been described as an “atmospheric pressure chemical ionization drift time spectrometer” (1-3) and a number of investigators have discussed the nature of the reactant gases and ion-molecule reactions involved (4-6). A good deal of uncertainty exists as to the exact identity of the product ion mass [M,H+, M(H2O)?H+,M,02-, etc.], and is compounded if solvents or impurities enter into the ion formation process. Spangler and Collins ( 4 ,Carroll et al. ( 5 ) , and Bird and Keller ( I ) discuss the nature of the ion-molecule reactions involved and suggest that the peaks are a composite of interacting species. Bird and Keller ( I ) discussed the dependence of the plasmagrams on sample vapor composition in their studies of “pulse” and “steady-state” sampling techniques. They discuss the overload pulse and “steadystate” sampling techniques. They describe the overload pulse sampling situation where an excess quantity of sample causes 0003-2700/79/0351-0870$01.OO/O
the reactant ion species to first disappear, then reappear as the sample concentration decreases with time. They approve of the pulse sampling overload situation on the condition that the resultant ion-molecule peaks are “time stable”. They contend that the peak is “time stable” if the peak maxima do not change by more than 0.2 ms during the lifetime of the peak. They propose that steady-state sampling is desirable to produce reproducible nontransient plasmagrams. This would of course be impossible for dynamic monitoring of peaks as they elute from a gas chromatograph. It has been our experience that reproducible plasmagrams are readily obtainable from GC peak sampling for a reasonable period of time where drift cell temperature and pressure are constant. The multiplicity and deformation of peaks often obtained can in some cases be explained by the presence of isomeric species and impurities. We have been able t o demonstrate that positional as well as stereometric isomers are resolvable with the GC-PC combination and are able to characterize, in one case, five separate positional isomers of a relatively high molecular weight compound. This does not mean that the plasma chromatograph in itself has high resolving power but it does point out the power the technique has in defining a chemical species with regard to the net effect of molecular ion mass and size. The size relationship is approximated using the average ion-neutral collision cross section for momentum transfer (0,) discussed and calculated experimentally by Lin et al. ( 7 ) . They utilize both static and rotational models of the ion to calculate a n average collision cross-section (Q,) value. This value is then used to calculate the reduced ion ) on a theoretical basis. They found that mobility ( K O term the mean cross-section values for the static and rotational models gave a satisfactory fit between experimental and 1979 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 51, NO. 7, JUNE 1979
theoretical predictions. Carr (8) has recently reported the characterization of ortho and meta dihalogenated benzenes utilizing the coated wire sampling technique and concluded that the difference in mobility between these isomers increases with the increased size of the constituent. Karasek et al. (3, 9) have studied isomeric halogenated nitro benzenes and underivatized phthalic acids via a combination of PC and MS techniques. Tou e t al. ( I O ) also have noted that plasma chromatography was isomer specific in their work with chlorodiphenyl oxides. An excellent review on the theory of plasma chromatography by Revercomb and Mason (11) deals with the average molecular cross section and we have used their derived equation with experimental ion mobility data to approximate differences in molecular cross section as related to isomeric species. The basic measurement is the drift time. which can be measured t o 0.02 ms on the apparatus as described. The reduced ion mobility value, KOis calculated from the equation,
1 t&
K--X-X-
273
P
T
760
where to = drift time, 1 = length of cell, E = field gradient, T = absolute temperature, and P = pressure. T h e reduced ion mobilities are correlatable between laboratories and are, of course, the principal form of data presentation. Revercomb and Mason (11) discuss the dependence of mobility on the ion-neutral collision and this dependence is expressed by the diffusion collision integral as average collision section RD shown in Equation 2.
where KO = reduced ion mobility (m2/V-s), = electronic charge (1.6021 X c), iV = gas number density (2.69 X lOI3/m3), m = mass of product ion (kg), M = mass of drift gas (kg), K = Boltzmann constant (1.3806 x IO-*, joules/K), T = absolute temperature (K),and RD = average collision cross section (m2). Rearranging Equation 2 allows us to use the experimentally determined K Ovalues to calculate the average collision cross sections. These values will be only as accurate as we can establish values for the mass of the product ion. More accurate measurements would be feasible if a mass spectrometer is utilized in conjunction with the drift tube but even known, the exact mass would be of limited value if the ion-molecule composition varies during transit in the drift tube. Griffin e t al. (6) discuss the mass-size relationship and point out that the direct mass dependence in the term [ l / m + 1/M]1!2 becomes insignificant as m gets much larger than M . I t is interesting however to compare the RD data for homologous series and in isomeric systems when we assume the product ions have equivalent transit compositions and degrees of reactant ion clustering. We recognize that the error in the collision cross-section value due to the uncertainty in molecular ion is unavoidable especially since the magnitude of error is also a function of the relative contribution between mass and molecular size in the determination of drift velocity. However, we assume a constant error for a given series and thus the cross-section values can be viewed for comparative purposes in the series even though the reduced ion mobilities are the prime form of data presentation.
EXPERIMENTAL Instrumentation. A Beta VI1 model plasma chromatograph from Franklin GNO Corp. (West Palm Beach, Fla.) equipped with a Nicolet Model 1072 signal computer was utilized to obtain the data reported in this work. Its mode of operation and functions have been described elsewhere (3,12-16). The reactant gas (flow rate, 100 mL/min) was cylinder air (high purity grade) and the
871
to FID
+
* l o P.C
. G.C
inlet via healed
line
Figure 1. The vake located in the GC oven is activated to select portions of the eluting component peak for PC analyses. In position a , the gas flows through the parts connected with the solid lines. I n position b, the gases flow through the parts connected with dashed lines. A Brooks flow controller (#8744)is used to control the N, flow from a tee in the N, supply to the PC unit
drift gas (flow rate, 500 mL/min) was from liquid nitrogen boil off. The instrument oven containing the drift cell was maintained at 230 “C for all of the data presented here. The high voltage was either positive or negative 3000 V for the particular ion polarity mode. Typical gate widths were 0.1 to 0.2 ms as noted with a 24-ms repetition period. The plasma chromatograms illustrated were plotted via the Nicolet and X-Y recorder system at 8 V full scale sensitivity with 0 delay and 20 ps /channel dwell time. The Nicolet has 1024 channels and the typical scan time is 20.48 ms; 1024 sweeps were used to obtain the data illustrated unless otherwise noted so peaks eluting from the GC were sampled for 21 s. The gas chromatograph was a Hewlett-Packard 5721 series equipped with an FID detector. Nitrogen carrier gas was used at a flow rate of 25 mL/min. A high temperature 4-port gas sampling valve (Valco v-4-ht Figure 1) was placed directly into the gas chromatographic oven with its shaft extending through the outside oven wall. The valve was used to selectively inject a slice of the chromatographic peak into the plasma chromatograph. This proved to be much more ideal than using the valving system as supplied in the GNO instrument where the valve is located in the PC oven. All of the chromatographic effluent must flow through the transfer line between the PC and GC in the latter case and resolution of gas chromatographic peaks and shoulders suffers. Column bleed tends to present a lesser problem also and “cleaner” plasma chromatograms are obtained when peaks are valved into the PC from the valve in the GC oven. In the normal position of the valve (solid lines), the column effluent is being diverted directly to the FID detector and the auxiliary make-up carrier gas of N2 (25 mL/min) is constantly sweeping through the transfer line and internal parts of the switching valve. The quantity of sample injected can be readily controlled by varying both the size of the GC sample injected and the length of time the valve is activated. When higher column temperatures are required, the “bleed” becomes more of a problem. Satisfactory plasmagrams are still obtainable by increasing the ratio of eluting component to the stationary phase bleed level. The transfer line (0.02-in. id., 1/16-in.0.d. stainless steel) from the GC is kept as short as possible and kept at 250 “C via a Variac and glass heating tape with appropriate insulation. Most of the gas chromatograms were obtained on a 6-ft. (1/8-in.0.d.) stainless column packed with 5% OV-101 on 80/100 mesh Chrom G-high performance. The perfluidone and methoxy substituted perfluidone samples discussed later were chromatographed on a 4-ft. (2-mm i.d.) glass column packed with 4% OV-101 on SO/lOO mesh Chrom G-high performance packing. On-column injection was utilized in all cases. The column oven was programmed from 100 to 300 “C at 10°/min. Normally, one plasmagram was obtained per GC run t o eliminate “clearing time” contamination between species. Materials. Isomer samples of fluorotoluene, dimethoxybenzene, picoline, quinoline, dipyridyl, triphenylene, chrysene, naphthacene, ethyl maleate, ethyl fumarate, and ethyl succinate were obtained from Alrich Chemical Company; acetotoluidides, Matheson, Coleman, and Bell; phenetoles, butyl maleate, butyl fumarate, toluic, phthalic, and succinic acids, Eastman Organic Chemicals. The esters of toluic, phthalic, and succinic acids were prepared via the BF,-catalyzed reagents from Pierce Chemical Corp. Methylated perfluidone and methoxy substituted perfluidone isomers were supplied by the Agrichemical Group - 3M Company.
872
ANALYTICAL CHEMISTRY, VOL. 51, NO. 7, JUNE 1979 TABLE I Isomer Reduced Ion-Mobility and Average Collisional Cross Section Ortho ___
KO
Meta
Para
n D
KO
n D
KO
+
ion
1.939
115.9
1.951
115.2
1.951
115.2
@)+
ion ion
1.733 1.738
127.0 126.7
1.761 1.757
125.0 125.3
1.766 1.763
124.7 124.8
ion
1.706
129.2
1.703
129.5
1.716
128.5
Toluic acid-methyl esters
+ +
ion -ion
1.680 1.691
130.2 129.3
1.649 1.663
132.6 131.5
1.671 1.674
130.9 130.6
Acetotolu idides
+
ion
1.623 1.610
134.8 135.9
1.602 1.587
136.6 137.9
1.614 1.599
135.5 136.8
1.540 1.527
138.5 140.6
1.450 1.481
148.0 145.0
1.466 1.497
146.5 143.5
Fluorotoluenes
a+
D i m e t h oxy benze nes Methyl phenetoles
-ion
+
Phthalic acid-methyl esters
ion
.ion
0 T h e original s a m p l e s were analysed via G . C . i n j e c t i o n of the peaks i n t o t h e P.C. w h i l e . 0 R e p r e s e n t s t h e same s a m p l e s i n j e c t e d in m e t h a n o l s o l v e n t directly i n t o the P.C. i n l e t three days later. K O is i n c m 2 / v o l t - s e cw h i l e
is expressed in A'.
Stationary phases and column support materials were obtained from Applied Science Laboratories. PM 0
RESULTS AND DISCUSSION One of the initial applications of plasma chromatography in our laboratory involved characterization of trace level impurities separated via gas chromatography. An effort was made to calibrate our instrument to obtain approximate molecular weight data. It was soon apparent that the multiplicity of peaks often obtained using the coated wire technique was a function of the purity of the compound involved. The valving systems described allowed us to take narrow slices of the GC peaks and it became quite obvious that positional isomerism was often responsible for distorted and multiple plasma chromatographic peaks. The gas chromatograph is an ideal sampling device allowing one to inject small quantities of relatively pure species without solvent clustering complications. A series of ortho, meta, and para isomers were then analyzed to establish the contribution of molecular configurations on reduced ion mobility. Table I illustrates the data obtained for a variety of compounds and, in most cases, the meta isomer exhibits the larger average collision cross section. Para isomers are intermediate and ortho isomers have the smaller cross-section values. Compounds with nitro, halogen, and hydroxyl substituents were avoided to minimize interpretation problems arising from dehalogenation, etc. in the ionization process. Figure 2 illustrates graphically the data for positive ions in Table I. As Carr (8)suggests, the differences in mobilities between isomers is a function of the size of the substitution constituent. Our results on the phthalate esters are not directly comparable to Karasek's work (9) in that he studied the free acids which were avoided in our work to minimize the memory effect of polar species on the walls of the cell. The ester data should be less prone to misinterpretation from dehydration reactions, etc. as discussed by Karasek where he reported MH+ as the most prominent ion for iso- and terephthalic acids while phthalic acid protonates and then dehydrates to yield the (M - HzO)H+ion. The data we have obtained indicate that the meta > para > ortho size relationship is most predictable for larger molecules even though the size of the constituent is small. The differences in the average collision cross-section values for isomeric systems may in part be correlatable to the dipole moments of the various species. Clustering charac-
POIM
OP
M
11
1
OP M
iI I
0
110
120
130 140 11- (average collision cross section)
150 i
p
Figure 2. Aromatic positional isomers (see Table I). Positive ions
teristics of the isomeric molecular ion with the reactant ion could be influenced by steric hindrance and relative basicity factors. Mass differences between isomers arising from variable degrees of reactant ion clustering have little effect on the RD value as calculated using the reduced mass term [ l / m + 1/M]'/' in Equation 2 . We found for example a difference of 6.4% in RD values for the ortho and meta phthalate esters. If we assume the molecular ion species of the meta isomer to be a simple protonation product [CloHloO,]H+,its RD value will equal 148.0 A'. If a four-unit water cluster associated with the molecular ion is assumed, [C10H1004](H20)4H+,its mass would be 37% larger and its RD value would be calculated as 145.5 A'. The difference in RD values for these two species would be 1.7%. The actual effect of the larger cluster would be t o increase the R D partially offsetting the decrease as calculated via Equation 2. We assume the M - H + species for the positive ion in this work in lieu of mass spectral information on degrees of clustering. I t should be emphasized that accurate data can be obtained on a short term basis where pressure and temperature are constant and the implied precision should be valid for comparison of the isomeric systems described. The average collision cross-section (QD)data are approximations since ionic masses are not accurately known. However, based on the assumptions previously mentioned, the comparative data are
ANALYTICAL CHEMISTRY, VOL. 5 1 , NO. 7, JUNE 1979 !lx128
Gas Chromatogram
I
1.024
199.3
202.4
I
J-(S0>Qh~”J
OCU,
n
1.040 Ch,
sGzcF,
(-JSOJ$;. SO,CF, CH,
Impurity Peak
1
IC
PLASMAGRAMS 18.56-
1.034
1.011
200.5
205.0
1.029
1.004
20i.4
ZOb.5
QW,QN:’“3
SO>CF, OCH,
CH,
(18.96
A
1.025
0 988
202.2
209.9
IIZ
PW
873
I
,
O
1,012
0.989
204.8 205
209.7 210
s4Q ~ \~”
I
SO,CF, CHI
200
s i D (average collision cross section) in i*
Flgure 3. Methylated perfluidone impurity. The upper chromatogram illustrates the impurity peak and the points at which the valve was actuated to obtain the plasmagrams. The solid lines represent the positive ions while the dashed lines represent the negative ions (ms) 1.063
Figure 5. Methoxy isomers of methylated perfluidone. The dashed lines are negative ions while the solid lines represent positive ions. The upper number is the K Oin cm2/V-s, while the lower number is the QD value in A*
1.031
1
..
201.8
195.7
& d H
y)
1.092
1,729
127.5
127.7
1.740
1.735
,
127.3
I
I
126.6
N’/q
1.732;
1.052 126
127
128
nD(average coliision cross “I
/cH, N
1 ,088
section) in A
2
Figure 6. Isomers of methylbenzotriazole. The dashed lines are negative ions while the solid lines represent the positive ions. The upper number is the K Ovalue in cm2/V-s and the Q, value is in A2
1.061
1.092
I
H Impurity
113.0 190.6 190
,2.000
197.6 192
194
196
198
20:
114.5
202
no (average collision cross section) in A’ l114.8
Figure 4. Characterization of methylated perfluidone impurity peak. The unknown impurity closely matches compound I1 as evidenced by GC retention time and both positive and negative reduced ion mobility data. The dashed lines are the negative ion while the s o l i line represents the positive ion. The upper number is the K Ovalue in cm2/V-s while the lower number is the 9, value in A2
of value and differences as small as 0.2 A2 can be measured. Figure 3 illustrates a practical example where an impurity was found in an experimental sample of methylated perfluidone. The impurity peak was partially isolated via the gas chromatograph and valved into the plasma chromatograph. T h e drift times in milliseconds are listed above the plasmagram peaks. The peak a t 19.46 ms was not identified or resolved by the gas chromatograph. Its chemical reactivity, GC retention time, and reduced ion mobility data indicate it is another isomer species. Reduced ion mobility data were thus obtained for the impurity and a series of reference isomers of perfluidone as shown in Figure 4. The unknown gave good positive and negative ion mobility spectra and was identified as compound 11. We subsequently isolated a sufficient quantity of this impurity via high performance liquid chromatography for NMR confirmation as compound 11. A second example of this type is illustrated in Figure 5 for five methoxy substituted positional isomers of perfluidone. Note that the meta-para-ortho positional effect appears to correlate to the size relationship discussed earlier for compounds 111, IV, and V. The differences between compounds I and I1 are more subtle and could be related to the sulfurphenyl ring bond. The combination of both positive and
1.782
I 124 2 11.773 1-24 9
1.671
130.5
1.575
I
110
i20 130 R , (average coi11sion cross seciionl
138.4 140 In H x
Figure 7. The upper numbers are K O in cm2/V-s while the lower numbers are 9, in A2 positive ions
negative ion mobility data and GC retention times yield positive component identification in this case. Figure 6 illustrates the data obtained for isomers of methyl benzotriazole. It is interesting to note in this case that the isomers have RD values which differ by less than 1 A2 and also the negative ions are very close in size to the positive ions. Isomers of picoline are illustrated in Figure 7 and it is interesting to note that the 2-picoline gives the smallest value for collision cross section while the 3- and 4-picoline isomers are nearly equivalent. Boiling point data and GC retention times also appear to correlate with the cross sectional values and may be of value in predicting the magnitude of isomer size differences. A different type of positional isomer influence on collision cross section is shown in Figure 7 . The 2,2’-dipyridyl yields a cross sectional value approximately 8 A* smaller than the
874
ANALYTICAL CHEMISTRY, VOL. 51, NO. 7, JUNE 1979
1'300
0
163.6
@ \
c11 C=c'
RO ti'
1.369
R = C.Hs
0
A I1
R = CaHn
1.616;
,1.577
,1.425
134.0'
137.2
'149.3
'OR
' H
0
Go
1.618,
,1.535 ;1.425
133.7
140.8 '149.2
1
11'257 169.3
'c- cn>-cti>-c, RO'
/
OR
155.4 0 /I
H\
Ro\C/C
1.375
/c,
= c\H
OR
1.618:
1.480,
J.425
133.8
146.3
149.3
130
150
T"OO 177.3
0
154.7
n,(average
172.2
160
170
~
collision cross section) in A I
Figure 8. The upper numbers are K O in cm2/V-s while the lower numbers are Q, in A* positive ions
4,4'-dipyridyl species. In this case, clustering characteristics of the ions involved may be playing a part in determining the relatively large difference in average collision cross-section values. The difference between quinoline and isoquinoline is much smaller but easily detectable. Ion mobility data of this type should be of value in assessment of bond order and molecular configuration. T h e isomeric condensed aromatic ring systems are illustrated in Figure 8. T h e species with the more linear configuration has the greatest average collisional cross section. This will be described more fully in a subsequent publication. Griffin et al. (6) have pointed out that polynuclear aromatic hydrocarbons should form MH+ ions from proton affinity considerations. They determined the ion formulas for a series of these aromatic compounds including chrysene which was investigated in this work. The ion formula and m / e were (C8HI2H)+and 229, respectively. We assume that the isomers naphthacene and triphenylene also form MH+ ions though we have no mass spectral data for confirmation on these two species. An example of cis-trans isomeric systems is shown in Figure 9. In this example, ethyl and butyl esters of maleic (cis) and fumaric (trans) acids were injected into the plasma chromatograph via the gas chromatographic interface. The corresponding esters of succinic acid were also examined to demonstrate the averaging effect of unrestricted rotation about the double bond. The average collision cross sections for the negative ions were equivalent for the particular ester species involved indicating that the double bonds are destroyed in the negative ion formation process. The positive ion data indicated that the more linear trans isomers were 8 t o 9 A* larger than the corresponding cis isomers. Succinic acid esters had intermediate cross sectional values even though their masses were two units greater than the corresponding species with restricted rotation about the double bond. Note that the differences between the esters of maleic and succinic acids are smaller than those differences between the fumaric and succinic acid esters.
CONCLUSION Plasma chromatography coupled with gas chromatography is a highly sensitive technique for the characterization of
n,
140 150 160 170 180 (average collision cross section) in 1 2
Figure 9. Cis-trans isomer systems. The esters of succinic acid which are 2 mass units higher in molecular weight still give intermediate values of 0, because of the averaging effect of rotation about the carbon-carbon bond. The dashed lines are negative ions while the solid lines represent positive ions
isomeric species. Average collision cross-section measurements provide insight in evaluating molecular configuration contributions of various functional groups. The data presented here illustrate that reduced ion-mobility measurements can readily differentiate between certain isomeric species. Peak broadening and multipeak plasmagrams can often be difficult to interpret without a knowlege of the isomeric structures present. Future studies should utilize a PC-MS system if possible, and compounds investigated should be chosen to illuminate the clustering and dissociative capture phenomenon which make interpretations difficult. The combination of gas chromatographic retention times and ion mobility data provides confirmational type data for the presence of trace level quantities of compounds which would be difficult to obtain with other techniques. Unknown compounds can be assigned only approximate molecular weights but relatively accurate ion mobility data for characterization purposes.
ACKNOWLEDGMENT The author would like to acknolwedge the assistance of A. Seaver, C. Green, R. Goff (3M Company), and H. E. Revercomb (University of Wisconsin - Madison). LITERATURE C I T E D (1) (2) (3) (4) (5) (6) (7) (8)
(9) (10) (1 1) (12) (13) (14) (15) (16)
G. M. Bird and R . A. Keller, J . Chromatog. Sci., 14, 574-7 (1976). R. A. Keller, A m . Lab., 7(5), 35-44 (1975). F. W. Karasek, Anal. Chem., 46, 710A-20A (1974). G. E. Spangler and C. J. Collins, Anal. Chem., 47, 393-402 (1975). 0. I . Carroll, I. Dzidic, R. N. Stillwater, and E. C.Horning, Anal. Chem., 47, 1956-9 (1975). G. W. Griffin, I . Dzidic, D. I. Carroll, R. N. Stillwell, and E. C. Horning. Anal. Chem.. 45. 1204-9 11973). S.N. Lin, C . W.Griffin. E. C: Horning, and W. E. Wentworth, J . Chem. f h y s . , 60, 4994 (1974). T. W. Carr, J . Chromatogr. Sci., 15, 85-8 (1977). F. W. Karasek and S.H. Kim, Anal. Chem., 47, 1166 (1975). J. C. Tou and G. U. Boggs, Anal. Chem., 48, 1351 (1976). H. E. Revercomb and E. A. Mason, Anal. Chem., 47, 970-83 (1975). F. W. Karasek and 0. S.Tatone, Anal. Chem., 44, 1758 (1972). F. W. Karasek, 0. S.Tatone, and D. M. Kane, Anal. Chem., 45, 1210 (1973). F. W. Karasek and D. M. Kane. Anal. Chem., 45, 576 (1973). F. W. Karasek and M. J. Cohen, J . Chromatogr. Sci., 9 , 390 (1971). F. W. Karasek and D. M. Kane, Anal. Chem., 46, 780 (1974).
RECEIVED for review September 8, 1978. Accepted March 19, 1979.
