sociate solute molecule, produces an increase in retention when k l and k-1 are small but has little, if any, effect on retention when k l and k-1 are large. The above analysis is based on the assumption that all reaction rate constants are known so that the elution peak can be simulated by means of a computer. However, in practice it is the peak shape which is known and the reaction rate constants are the sought-for information. In principle, all on-column kinetics can be obtained from an analysis of the peak characteristics. However, this is not an easy task. Work is continuing along these lines in our laboratory.
ACKNOWLEDGMENT
We wish to thank S. H. Lin for suggesting the use of the steepest descent approximation method in the solution of this problem and Joseph Fulton for programming the plotter package used in the preparation of Figures 2-4. Received for review January 7, 1974. Accepted April 19, 1974. Financial support from the National Science Foundation under grant GP25553 and from the National Defense Education Act for a Fellowship for one of us (D.L.D.) is also gratefully acknowledged.
Determination of Ionization Potentials by Gas-Liquid Chromatography Richard J. Laub and Robert L. Pecsok Department of Chemistry, University of Hawaii, Honolulu, Ha waii 96822
The possibility of determining vertical ionization potentials by GLC is explored. Charge transfer formation constants (K1)were determined for benzene, toluene, and the xylenes with 2,3-dichloro-5,6-dicyanobenzoquinone (DDQ) in di-nbutyl phthalate (DNBP). These were found to vary inversely with ionization potential. Flfteen substituted butadienes were then examined with 2,4,7-trinitrofluorenone (TNF) in DNBP. Several of these yielded abnormally low formation constants, resulting in high apparent ionization potentials. Steric hindrance (mutual approachability) is postulated as a limitation of the method. Other ionization potentials determined by this technique are shown to be in excellent agreement with predicted values.
Gas-liquid chromatography (GLC) has been used in conjunction with charge transfer complexes for some time (13). Purnell ( 4 ) has reviewed various classes of solute-solvent interactions. Gil-Av and Herling ( 5 ) ,and Muhs and Weiss ( 3 ) have proposed a method of calculating the charge-transfer formation constant, Kf. For D A s C,
+
and
where D, A, and C represent donor, acceptor, and complex species, respectively. KLO is the distribution coefficient for a donor species on a column containing only “inert” stationary phase. KL is the distribution coefficient on a column with acceptor concentration [A] dissolved in the liquid phase. Other methods of determining Kf by GLC have also been proposed (6). R. 0. C. Norman, Proc. Chem. Soc.,151 (1958). E. Gil-Av, J. Herling, and J. Shabtai, J. Chromatogr., 1, 508 (1958). 84, 4697 (1962). M. A. Muhs and D. T. Weiss, J. Amer. Chem. SOC., J. H. Purnell, in “Gas Chromatography-1966,’’ A. B. Littlewood, Ed., Adlard, London, 1966, pp 3-20. (5) E. Gil-Av and J. Herling, J. Phys. Chem., 66, 1208 (1962). (6) D. E. Martire and P. Riedl, J. Phys. Chem., 66, 1208 (1962).
(1) (2) (3) (4)
1214
Kf is related to the true thermodynamic equilibrium constant, Keq,by:
K , = Kf
YCa
(3 )
YBaYAa
where yimis the infinite dilution activity coefficient of the ith species. Eon and Karger (7) have recently calculated “corrected” and “relative absolute thermodynamic” formation constants from Purnell’s excellent data (8).These were for toluene, ethylbenzene, and the three xylenes, with T N F in a variety of alkyl phthalate ester solvents. If true charge transfer forces are operative, one would expect the measured formation constant to be a function of the ionization potential (Id)of the donor, and the electron affinity ( E a )of the acceptor. If solvent interactions are negligible, or can be taken into account, this is approximately true. When large solvent effects occur, no correlations between Kf, Id,or Ea are expected. This is apparent in spectroscopy where the Benesi-Hildebrand equation (9) (or various modifications) is used to determine the formation constant. Anomalies are amply illustrated by Foster and Fyfe (10); and Brown, Foster, and Fyfe ( 1 1 ) who employed proton NMR and 19FNMR spectroscopy. The latter technique particularly would be expected to overcome solution effects. They found, however, that the use of different solvents made it impossible to compare either Kf values, or the ratios of such values. When log Kf (1,3,5-trinitrobenzene and fluoranil) was plotted against Idfor 26 alkylbenzenes, the correlation was poor (12). The same lack of linearity was noted for -AHp us. I d (13).Dewar and Thompson (14) have severely criticized charge transfer theory, partic(7) C. Eon and B. L. Karger, J. Chromatogr. Sci., 10, 140 (1972). (8) D. L. Meen, F. Morris, and J. H. Purnell, J. Chromatogr. Sci. 9, 281 (1971). (9) H. A. Benesi and J. H. Hildebrand, J. Amer. Chem. SOC., 71, 2703 (1949). (10) R. Foster and C. A. Fyfe, Trans. Faraday SOC.,62, 1400 (1966). (11) N. M. D.Brown, R. Foster, and C. A. Fyfe, J. Chem. SOC.B, 406 (1967). (12) P. H. Emslie, R. Foster, I. Horrnan, J. W. Morris, and D. R. Twiselton, J. Chem. SOC.B, 1 16 1 (1969). (13) M. I. Foreman, R. Foster, and C. A. Fyfe, J. Chem. SOC.B, 528 (1970). (14) M. J. S. Dewar and C. C. Thompson, Tetrahedron SuppL, 7, 97 (1966).
A N A L Y T I C A L C H E M I S T R Y , VOL. 46, NO. 9, AUGUST 1974
ularly as proposed by Mulliken (15). Foster's data clearly indicate Dewar was correct in being suspicious of relations between Kf and Idor Ea determined spectroscopically. Solvent and absorption variations, temperature, and differences in purity make such correlations difficult. The determination of formation constants by GLC is not so constrained, however, as Martire has pointed out (6, 16). KLO in Equation 1 is unambiguously found by chromatographing the donor on pure stationary phase. The concentration of acceptor is known within experimental weighing error (generally tenths of a milligram), leaving Kf to be calculated from the slope of KL us. [A] plots. The accuracy of such GLC determinations can be seen in Reference 8. Purne11 and Srivastava ( 17) have very recently confirmed the advantage of the GLC technique. They also demonstrated that, for their experiments, NMR and UV-determined formation constants were not meaningful. Undoubtedly, solvent effects are very important in spectroscopic studies, overriding charge transfer forces. The GLC method, however, was shown to yield useful formation constant data. Hence, Kf-Zd correlations may now be examined using gas-liquid chromatography. Srivastava and Prasad (18, 19) have found that, in general, donors with lower ionization potentials yield higher formation constants when measured spectroscopically. Others have noted similar trends (20-22). Many authors have reported a linear correspondence of Idto the energy of charge transfer, ECt(23). Rose (22) has suggested the use of charge transfer absorption frequencies to determine ionization potentials, first postulated by McConnell (24) in 1953. Such studies will be valid, however, only for similar donors with the same acceptor in the same solvent. If xylenes are compared to polycyclic aromatic hydrocarbons, for example, no correlation is apparent or expected. Our exploratory investigations have examined the possibility of determining ionization potentials from formation constants by GLC. Simply, known ionization potentials of closely-related donors are plotted against respective formation constants determined from GLC data. The Idof other similar donors whose Kf values have simultaneously been determined, can then be found. EXPERIMENTAL Apparatus. A Hewlett-Packard Model 402 dual flame gas chromatograph was employed. T h e recorder was a Varian A-25. A Vidar Model 6230 Digital Integrator was used as a stopclock for aromatic hydrocarbons. The integrator was continually checked against the recorder. Flow rates were determined by timing the rise of a soap-bubble in a water-jacketed calibrated tube with an Industrial Timer Corp. Model SC-100 stopclock capable of %oo-sec accuracy. T h e column temperature was controlled by jacketing the U-tube and pumping thermostated water from a 5-gal insulated reservoir through the jacket. Temperature of the water was maintained to *0.02 "C, measured a t the jacket entrance with measuring and reference (ice-water) iron-constantan thermocouples connected to a Leeds and Northrup Type K-3 potentiometer. A Keithley Instrument Model 153 microvolt-ammeter was used as a center-scale galvanometer. A Tronac Model PTC-1000 Precision Temperature Controller was employed with 250-watt and 800-watt (Variac-controlled)immersion heaters for the water reservoir. (15) R . S. Mulliken, J. Chem. Phys., 61, 20 (1964). (16) J. P. Sheridan, D. E. Martire, and Y . B. Tewari, J. Amer. Chem. Soc., 94, 3294 (1972). (17) J. H. Purnell and 0. P. Srivastava, Anal. Chem., 45, 1111 (1973). (18) R . D. Srivastava and G. Prasad, Spectrochim. Acta, 22, 1969 (1966). (19) R. D.Srivastava and G. Prasad, Bull. Chem. Soc. Jap., 43, 161 1 (1970). (20) L. J. Andrews, Chem. Rev., 54, 713 (1954). (21) J. N. Murrell Quart. Rev., 15, 191 (1961). (22) J. Rose, "Molecular Complexes." Pergamon Press, Oxtord, 1967 (and
references therein). Mulliken and W. B. Person, "Molecular Complexes." Wiley-lnterscience, New York, N.Y., 1969. (24) H. McConnell, J. S. Ham, and J. R. Platt, J. Chem. Pbys., 21, 66 (1953). (23) R . S.
Reagents. Dienes and aromatic hydrocarbons were obtained from Chemical Samples Co., Phillips Petroleum, or Aldrich Chemical Co. Di-n-butyl phthalate was reagent grade from Kodak and was redistilled through a Podbielniak column (bp, 151.5-152.0 "C a t 0.9 Torr). DDQ and T N F (both from Kodak) were each twice recrystallized from anhydrous benzene, mp, 214 "C and 176.5 "C, respectively. Gas Chrom Q (100/120 mesh) was used as the solid support for aromatic hydrocarbons and Chromosorb W (60/80 mesh, AW-DMCS treated) for the diene experiments. Each solid support was previously treated with hexamethyldisilazane (HMDS) from Bio-Rad Laboratories. Procedure. Column packings were made by slurrying carefully weighed amounts of acceptor, stationary phase, and sieved support in a volatile solvent (anhydrous benzene for DDQ; acetone for T N F ) , and stripping off the solvent by slow rotary evaporation just until clumping occurred. T h e packings were then subsequently dried in a fluidized bed apparatus a t 60 "C. Each was then packed in the 5-ft by '/4-in. jacketed U-tube which had previously been HMDS-treated. Very gentle tapping was used to introduce the packing, after which a small gas pressure and gentle tapping on the sides of the column were used until no more packing could be added below Yd-in. from the top of each side. Silanized glass wool plugs were then inserted into each end, care being taken not t o crush the solid support. Conditioning was a t 60 "C for a t least 5 hr a t 20 ml/min He flow rate. Densities of the stationary phase-complexing agent solutions were determined as a function of temperature by measuring the rise of a known weight of solution in a calibrated capillary tube thermostated to the temperature of interest. The amount of stationary phase on each packing was determined by carefully emptying and weighing the column contents. Known amounts were then ashed. Liquid phase and complexing agent weights could then be calculated assuming negligible transfer losses in making the packing. Samples were injected neat or as well-resolved mixtures, no differences in retention times being noted. A Hamilton 1-11 syringe was rinsed with solute, the excess expelled, and the remainder (the amount coating the inner surface of the needle, about 0.005 pl) was injected on-column as vapor. Column dead volumes were measured by injecting methane. The column head pressure was determined by injecting a needle connected to an open-end manometer through the septum. OV-7 on Chromosorb W was used as a reference column. No variation of retention time with sample size was seen from less than 0.005 pl t o 0.04 pl. All peaks exhibited Gaussian shape, and all data were reduced by algorithms written in PL/1 on an IBM 360/65 computer. Distribution coefficients were calculated as usual from the fully corrected net retention volume and column liquid volume. Formation constants were then determined as the slope/intercept quotient of the least squares line formed by K L us. [A] plots. The observed us. calculated KLO values agreed to no worse than 20/, and usually better than 1%.Cyclohexane was routinely eluted as an internal standard and gave zero slope by the above procedure to better than 0.010 ]./mole.
RESULTS AND DISCUSSION S t a n d a r d Ionization Potentials. Ionization potentials determined from formation constants by this GLC method are dependent on the "standard" Idvalues from the literature, used in defining KpZd plots. Mulliken (23) and others have consistently pointed out that vertical as opposed to adiabatic ionization potentials must be used. A vertical value, Zvd, is the ionization potential for a donor which retains its groundstate geometrical configuration after ionization has occurred. Photoelectron spectroscopy (PES) data are now generally recognized as an accurate measure of ZVd values; Table I lists the average PES Zvd literature values for aromatic hydrocarbon donors in this study. Table I1 contains Zvd literature values for the dienes. Aromatic Hydrocarbons. The KLO and Kf values of Purnell's work (8) determined for aromatic hydrocarbons with T N F in di-n-propyl phthalate (DNPP) at 60 "C are presented in Table I. Also given are the values obtained by this investigation at the same temperature, using DDQ and DNBP. Our KLO values are somewhat higher than those of Purnell, which may be due to the employment of different
A N A L Y T I C A L C H E M I S T R Y , VOL. 46, NO. 9, A U G U S T 1974
1215
“standard” literature Evavalues must be available. Briegleb (33) has calculated Eva for 74 acceptors, relative to chloranil. The absolute values were f0.2 eV a t best, but the relative order and magnitudes were acceptable compared t o other methods. An interesting class of compounds we are now examining, for example, are the pyridazinediones (34, 35),for which no electron affinity data exist. Ethylbenzene gives a much lower formation constant with T N F than expected, based on ionization potentials (ethylbenzene was not available to us a t the time this work was carried out). Steric hindrance has long been recognized as an important factor in charge transfer interactions (20), and may explain this anomaly. Martire (16) has noted a
Table I. Vertical Ionization Potentials (25-28), Distribution Coefficients, and Formation C o n s t a n t s for Aromatic Hydrocarbons w i t h TNF and DDQ at 60 “C T N F (8) Aromatic
Av. Ivd, eV
Benzene 9.25 Toluene 8.82 Ethylbenzene 8.76 m-X ylene 8.57 8.57 o-Xylene p-Xylene 8.46
Kf, 1.imole
K L ~
210 533 1159 1253 1604 1210
, , ,
0.237 0.161 0.281 0.319 0.331
DDQ K L ~
286 707 ... 1704 2174 1662
Kr, l./mole
0.849 0.930 ... 1.060 1.058 1.090
Table 11.Vertical Ionization Potentials (28-32), Distribution Coefficients, and Formation C o n s t a n t s for Dienes w i t h T N F Diene No.
Name
1 cis-1,3-Pentadiene 2 2,3-Dimethyl-1,3-butadiene 3 trans,trans-2,4-Hexadiene 4 trans,trans-2,5-Dimethyl-2,4-hexadiene 5 2-Methyl-1,3-butadiene 6 1,3-Cyclohexadiene 7 4-Methyl-1,3-pentadiene 8 2-Ethyl-1,3-butadiene 9 1,3-Hexadiene
10 2,4-Dimethyl-1,3-pentadiene 11 2-Methyl-1,3-pentadiene 12 3-1Clethyl-1,3-pentadiene
13 14 15 16
5-Methyl-1,3-hexadiene 1,3-Heptadiene 2,4-Heptadiene l-Methoxy-1,3-butadiene
8.67 8.70 8.33 7.88 8.87 8.25 ... ...
... ... ... ... ... ... ... ...
45=
500
550
450
50’
89.69 183.2 286.3 207.4 62.68 332.9 252.2 167.3 206 .0 208.4 231.3 269.3 345.5 504.7 704.4 602.9
78.11 156.5 240 .0 165.5 54.49 281.1 214.3 142.2 174.6 177.3 196.3 227.7 290.6 420.3 580.0 499.3
66.89 131.9 201.1 133.4 47.48 238 .O 178.6 121.1 146.3 148.6 165.9 189.8 241 .O 345.1 473.2 412.2
0.078 0.066 0,143 0.192
0.057 0.054 0.120 0,181 0.008 0.128 0.096 0.038 0.051 0.006 0.087 0.090 0.018 0.042 0.049 0.167
“inert” solvents (bp DNPP, 304O, bp DNBP, 340’). When the ratio of KLO values from Purnell’s data to our work is multiplied by our donor formation constants, the resultant “normalized” formation constants for DDQ are two to three times those obtained with TNF. DDQ serves as a good contrast to TNF: it is freely soluble in dialkyl phthalates, and has an electron affinity of 1.96 eV, compared to 1.00 eV for TNF (33).If true charge transfer forces are operative in these systems, one would therefore expect formation constants with DDQ to be about double those for T N F (as our results indicate), although KPE,,~relations may not necessarily be linear (22). Nevertheless, our data point to the possibility of determining vertical electron affinities by GLC. Formation constants for the same donors with several different acceptors would be plotted against the known Evavalue for each acceptor; new electron affinities could then be found by determining the donor formation constants on acceptors with unknown Evavalues, and referral to the Kf-EVagraph. As with ionization potential determinations, however, no steric effects may be present, and (25) K . Watanabe, J. Chem. Phys., 22, 1564 (1954). (26) K. Watanabe, T. Nakayama, and J. Motti, J. Ouart. Spectrosc. Radiat. Transfer,2 , 369 (1962). (27) R. Braisford, R. V. Harris, and W. C. Price, Proc. Roy. Soc., Ser. A, 258, 459 (1960). (28) D. A. Demeo and M. A. El-Sayed, J. Chem. Phys., 5 2 , 2622 (1970). (29) J. L. Franklin, J. G. Dillard, H. M. Rosenstock, J. T. Herron, K. Droxyl. and F. H. Field, “ionization Potentials, Appearance Potentials, and Heats of Formation of Gaseous Positive Ions,” NSRDS-NBS, No. 26, National Bureau of Standards, Washington, D.C., 1969. (30) D. A. Labianca, G. N. Taylor, and G. S. Hamrnond, J. Amer. Chem. Soc., 94, 3679 (1972). (31) W. C. Price. R. Braisford, P. V. Harris, and R. G. Ridley, Spectrochim. Acta, 14, 45 (1959). (32) J. L. Franklin and A. Mogenis, J. Phys. Chem., 71, 2820 (1967). (33) G. Briegleb, Angew. Chem. Int. Ed. Engl., 3 , 617 (1964).
1216
Kr, l./mole
KLQ
Av. IVd, eV
0.015
0,155 0.107 0.045 0.062 0.021 0.096 0.102 0.035 0.047 0.057 0.170
550
0.048 0.047 0.088 0.151 0.007 0.102 0.080 0.030 0.047 -0.001 0.076 0.079 0.015
0.033 0.045 0,143
similar problem with aromatic hydrocarbons and tetra-nbutyl pyromellitate. The ester groups of this acceptor are free to rotate out of the plane of the phenyl ring. Donors are thus prevented from approaching the acceptor closely, yielding the anomalous order of formation constants: benzene > toluene > o-xylene > p-xylene > m-xylene > mesitylene. Similar effects have also been found in spectroscopic work (12),although these latter data may not be definitive for reasons given above. This indicates a limitation of our GLC method: there must be no steric hindrance or out-of-plane deformations by either donor or acceptor. Where such factors occur, lower formation constants (hence higher apparent ionization potentials) will be obtained. The Kf values for m- and o-xylene with T N F are not in complete agreement in Table I. However, assuming that 0.30 l./mole is the correct value for both (the average of the two), the percent error is negligible in Purnell’s data. T o test whether or not the order of formation constants was fortuitous, we therefore employed DDQ, a much stronger complexing agent in a slightly different stationary phase. The formation constants for this complexor in Table I proceed as expected: lower ionization potentials yield higher formation constants. The Kf values for o- and m-xylene are virtually identical, as are the known ionization potentials. An additional point to be examined is the separation of m- and p-xylene on DDQ. Purnell ( 4 , 8 ) has derived the general expression for a solute pair:
(4) (34) R. H. Mizzoni and P. E. Spoerri, J. Amer. Chem. Soc., 7 6 , 2201 (1954). (35) K . Eickenberger, A. Stachelin, and J. Druey. Helv. Chim. Acta, 37, 837 (1954).
A N A L Y T I C A L C H E M I S T R Y , VOL. 46, N O . 9, AUGUST 1974
Table V. Steric Effects and Apparent Ionization Potentials
Table 111. Comparison of Known and Calculated Ionization Potentials Diene No.
Av. I T d ,eV (Fig. 1)
I , d , e V (PES)
1 2 3 4 5 6
8.65 8.66 8.27 7.88 8.96 8.20
8.67 8.70 8 33 7.88 8.87 8.25
yo Error
Diene No.
0.23 0.46 0.72 0.00 1.02 0.61
I = / = /
15 9 14 8
Table IV. Apparent Ionization Potentials
13
I , - d , eV (from Figure 1)
Diene No.
7 8 9 10 11 12 13 14 15 16
45
3c
8.54 8.77 8.72 8.84 8.59 8.56 8.80 8.77 8.74 8.12
50 ' C
55 " C
Av. I v d ,eV
8.48 8.75 8.70 8.85 8.53 8.51 8.82 8.74 8.71 7.98
8.45 8.75 8.67 8.85 8.48 8.45 8.80 8.74 8.68 7.94
8.49 8.76 8.70 8.85 8.53 8.51 8.81 8.75 8.71 7.98
Structure
I:,
e\'
8.67 8.71
8.70 8.75
7 '
8.76
7
8.81 8.83
Table VI. Vertical Ionization Potentials by Gas--Liquid Chromatography Diene No.
where K+ is the formation constant of a given solute, LY is the relative volatility of the donor pair for the stationary phase alone, and p is the relative volatility of the pair for the stationary phase containing an acceptor concentration, [A]. Setting @ = 1.25 (corresponding to a column of about 1000 theoretical plates for separation), and CY = 1.03 (about the same for T N F and DDQ), K f / K f b must be 1.29 before separation is achieved. According to Table VII, Ref. 8, Kfb must hence be about 10 l./mole a t [A] = 5M for resolution of the pair. This is unrealistic, even for DDQ, which presumably is the strongest complexing agent available (based on electron affinities), since Kf is on the order of 1 l./mole at 60 "C. Kf is larger a t lower temperatures but the volatility of aromatic hydrocarbons simultaneously decreases, greatly increasing analysis time. Obviously, higher molecular-weight species, such as polycyclic aromatic hydrocarbons, will not elute a t all a t such low temperatures. Highspeed liquid-liquid chromatography (LLC) may prove useful in this regard. Formation constants can undoubtedly be determined by LLC but may be plagued with solvent effects, as in spectroscopic methods. For purposes of separation and identification, however, these effects may be adjusted to yield relatively large differences in solute retention. For example, Srivastava and Prasad (19) have reported spectroscopic Kt values of 2.50 and 3.00 l./mole for m- and p-xylene, respectively, a t 25 "C in CHC13. While these values may be erroneous when compared to GLC Kf values, nevertheless, they may prove entirely satisfactory for quantitative and qualitative purposes. The Kf ratio approaches the necessary value of 1.29, and may be adjusted to yield even larger (although erroneous) values merely by varying the mobile phase. The further possibility of changing the formation constants during elution exists with gradient capabilities in LLC. For example, the separation and identification of polycyclic aromatic hydrocarbons is of great practical importance in air pollution monitoring. Formation constant alteration during LLC analysis of such compounds may therefore prove to be useful in this regard. Dienes. Diels-Alder reactions occur with dienes and DDQ, hence T N F in DNBP was used as the complexing system. The relevant data are presented in Table 11, and
Structure
l " d , eV
11
8.53
12
8.51
7
8.49
16
w d C H 7.98
plotted in Figure 1. To ascertain the accuracy of the equations given in the figure, the ionization potentials for dienes 1-6 were back-calculated, and are compared to the values in Table 111. The averaged values agree known IVd with the known I,dvalues to better than 1.02%, and the average error is 0.51%. There is therefore no doubt about the curvature of Kf us. Ivdplots f o r these experiments, even though the measured formation constants border the experimental error of the method (ca. 0.010 l./mole). The apparent diene ionization potentials from Figure 1 are given in Table IV. Steric effects are immediately obvious as shown in Table V. As alkyl substituents on the butadiene skeleton become bulkier, the formation constants decrease. This leads to higher calculated ionization potentials. 2,4Dimethyl-l&pentadiene (No. 10) may be deformed out of plane by methyl-methyl repulsions. Conjugation would then be partially destroyed. This seems to result in a larger decrease in Kf than steric effects. Further research with sterically hindered and partially deconjugated systems is obviously indicated. Table VI lists vertical ionization potentials which are not sterically hindered. These should therefore be valid to h0.1 eV (1-2% experimental error on Kf). A nonlinear least squares program ( P L / l ) was used to determine the equation of each curve in Figure 1 (36, 37), where the general equation, Y = a x 2 -t b, was employed since the curves appeared to be parabolic. The equations given in Figure 1 fit the data to 2% in all cases, and usually better than 1%,The significance of these parabolic curves is not immediately apparent, although linear relations between donor ionization potentials and resultant formation (36) W. E. Wentworth, J. Chem. Educ., 42, 96 (1965). (37) W. E.Wentworth, J. Chem. Educ., 42, 162 (1965).
A N A L Y T I C A L C H E M I S T R Y . V O L . 46, NO. 9, A U G U S T 1974
1217
K , = C31,d2
0.2 0
+
C4
(6 1
Kf and ECtare then empirically related by Equations 5 and 6. Usefulness of the Method. We have shown that GLC can be used to determine vertical ionization potentials where no anomalous (e.g., steric hindrance) effects occur. The values obtained are accurate to f0.1 eV. This was illustrated in Table 111, where known ionization potentials were compared to the average of calculated values determined from the respective formation constants and equations in Figure l. I t should also be possible to determine electron affinities by the same method. Of potentially greater usefulness, however, is the unambiguous determination of formation constants, steric effects, and the partial deconjugation of systems. For example, we are now examining sterically hindered systems in an effort to quantify (at least on a relative basis) out-of-plane deformation angles.
0.15
5-
0.10
Y
CONCLUSIONS
0.05
0.00
7
Figure 1. Plot of Kfvs. .Id. The approximate equation constants are:
+
45 ‘ C : 4 = -9.075 X Ld2 0.750 50 ‘ C : 6 = -9.237 X loT3 kd2-k 0.750 55 ‘ C : & = -9.445 X hd2 0.750
+
constants probably should not be expected. I t is interesting to note that the energy of charge transfer, ECt,has been shown (22, 23) to fit the following empirical equation:
where C, is an empirical constant. In this research,
A final point to be made regards the nature of “charge transfer interactions.” These interactions may involve actual complex formation, loosely-bound contact pairing, or complexation of some intermediate strength (23).A further complication arises, since most of the available literature data has been obtained spectroscopically, yet Purnell has recently demonstrated that such methods yield questionable results (17). These and other disagreements lead the authors to speculate that what is presently termed “charge transfer” is not well understood. Such interactions may furthermore only be a reflection of solution phenomena which have yet to be elucidated. ACKNOWLEDGMENT
The authors acknowledge A. F. Isbell, Jr., R. S. H. Liu, and V. Ramamurthy for many helpful discussions and most of the dienes. We also very gratefully thank R. L. H. Williams for his Job-like patience, and glass-blowing skills. Received for review October 18, 1973. Accepted April 18, 1974.
Gas Chromatographic Detection and Confirmation of Volatile Boron Hydrides at Trace Levels E. J. Sowinski Western Electric Company, Allentown, Pa. 18103
I. H. Suffet’ Department of Chemistry and Environmental Engmeering and Science Program, Drexel Unlverslty. Philadelphia. Pa. 19104
A characterization of gas chromatographic detectors is presented in a plan which provides for airborne identification of specific boron hydrides of significance as environmental contaminants. Lower limits of detectability for electron capture, microcoulometry, and flame photomet-
To whom reprint requests should be sent. 1218
ric detection are described at ppb levels which are in the range of industrial air standards (Threshold Limit Values). Interferences of significance to environmental analysis are described and a GC-MS interface for confirming boron hydrides is presented. The advantages of combining GC detectors in environmental analysis for boron hydrides are discussed.
A N A L Y T I C A L CHEMISTRY, V O L . 46, NO. 9. AUGUST 1 9 7 4