Table I.
Relative Retention Time
(2,g-xylidene Stationar) liquid LVt c c Celite 34.3 mesh Column length, meters Column temp , C Carrier gas
Silicone grease 10 80-100 3 75 155
s*
1 00
1 (10
TXPa 10 80-100 2 25 127 ?;,
1 00 1 04
=
1.00)
PEG 4000 10 80-100 2 25 128
Sorbitol 10 60-80 3 75 127 5
s,
1 00 1 19
DBS 10 40-60 2 25
Ii2
130 He
1 00 1 14
1 00 1 13
DBS 15, PEG 4000 0 5 60-80 2 25 130 He 1 00 1 1.5
DBS 10, DGO 4 60-80 2 25 130 He 1.00 1.14 1.22 1.34 1.48 1.63
2,3-Sylidine 3,4-?(ylidine T I P = Tri-2,4-uylenyI-phosphate
Table II.
Determination of Isomeric Xylidines-
Xylidines, 70 Sample 2,6- 4 2$2,43,sCommercial xylidine mixture 14.0 18.3 44.0 3.0 Xylidine fraction in coal tar 1.7 15.7 26.2 48.4 e Peak area determined directly from the integrated values using the the peaks as a dividing line.
RESULTS A N D DISCUSSION
Results for several liquid I,hases are shown in Table I. The best separation was ohtained using DBS-DG column ]lacking as shown in Figure 1. K h e n the amount of I>G or PEG as tailing reducer was increased, 2,5-xylidine and 2,4-
2,3-
324
13.3
7.3
2.0 6.0 minimum between
xylidine elut'ed simultaneously, but 2,4xylidine was thoroughly separated from 3,5-xylidine. The analytical utility of t,his DBS column is demonst'rated by the data in Figure 2 which show the chromat,ogranis of a commercial xylidine mixture and a
xylidine fraction in coal tar. The semiquantitative analyses of xylidine isomers by the direct area normalization method are shown in Table 11. LITERATURE CITED
(1) Fitzgerald, J. S.,Australian J . ilppl. Sci. 12, 51 (1961). (2) James, A. T., ANAL.CHEM.2 8 , 1565
(1956). (3) Jones, J. H., Iiitchie, C. D., Heine, K. S., J . Assoc. Ofic. Agr. Chemasts 41,749 (1958). W A T A R C FUNAs.4KA
TSUGIO KOJIMA HIROYUKI IGAKI Faculty of Engineering Department of Industrial Chemistry Kyoto University Yoshida, Kyoto, Japan
Quantitative Infrared Microanalysis of High-Boiling Aliphatic Neutral Oil Fractions SIR: In characterizing neutral oils from low-temperature coal tars a quantitative infrared microanalytical procedure had to be developed that could be applied to undiluted samples of aliphatic$ with a variety of both saturate and olcfin types in the CISto C?o range, inclusive. -1 survey of the literature showed that a procedure for this molecular weight range and class misture had not bwn described. Saturates and o1tfm.i have been dealt with separately and the work has been either on relatively low-boiling material or polymers. In othrr work the analyses were cond w t e d in solution and the use of ultramicrocavity cells with a beam condenser for surh an analysis has not been dewri bcd.
pure grade hydrocarbons, in so far as these were available. Apparatus. Spect'ra were recorded on a Perkin-Elmer Model 21 infrared spectrophotometer equipped wit'h sodium chloride optics and a PerkinElmer 6s ultramicrosampling unit mounted in t h e sample beam. With proper alignment, the reflecting optics of the beam condenser will t,ransmit about 40y0 of the energy normally available. The cell masks, with dimensions of 1 X 4 mm., do not further reduce the energy. Sodium chloride liquid ultramicrocavity cells (Type I), Connecticut Instrument Corp.) with nominal pathlengt,hs of 0.05 mm. were used to obtain spect'ra of both pure compounds and neutral oil samples. The liauid cell holder has been described * ( 5 ) . Procedure. The neutral oil dis...~ tillate fractions analyzed in this work were obtained from a low-temperature bituminous coal tar. These had been separated into aromatics and aliphatics bj- countercurrent dist,ribution in a dual solvent system of isooctane and ~
EXPERIMENTAL
Reagents. The pure compounds u w l in the calibration ITere either .A me ri c a n Pet ro le u in I n ?t i t u t e c. t andard xanil)le.: or Phillips Petroleum Co.
~
~~
90 wt. % ethanol in water. T h e aliphatic material was recovered from the upper phase solvent (isooctane) in the automatic fraction collector of a Craig countercurrent distribution (CCD) apparatus. The material was separated from aromatics except for an overlap of a few tubes. However, only a slight separation of saturates and olefins by class mas effected (about two tubes). I t wab necessary t o use ultramicrocavity liquid cells and a beam condenser to obtain infrared spectra because the amount of sample was often as low as 2 me;. after so1 .ent removal. The analysis was performed in a solventfree state because of the low absorptivities of aliphatic compounds and also t o avoid solvent interference at the analytical wavelengths. For the pure compounds used in setting up the analysis, molar absorptivities (e) a t the analytical wavelengths were determined from the relationship e = .1.1f/1000 b d 2 5 (1) where A is absorbance, b is cell thickness in centimeters, d 2 j is the density of VOL. 36, NO. 1 1 , OCTOBER 1964
2215
the pure sample a t 25" C., and M is the molecular weight of the pure sample. For pure a-olefins in the undiluted state, t h e absorption band a t 11.02 microns reached 100% absorption in a 0.05-mm. cell. Therefore, it was necessary to use the solution-determined value for this band. This was considered justifiable because the molar absorptivity determined in the undiluted state for the 6.07-micron a-olefin band (which did not reach 100% absorption in a 0.05-mm. Fell) agreed with that determined in solution. Because the neutral oil samples were mixtures, their bands remained on scale in the undiluted state in 0.05-mm. cells and absorbances were thus measured directly.
5000 s100. ' 2
- 80
1
'
' L ' ' ! ' '
NO.
range Ca-Cu
2-Me thylalkanes trans-Internal olefins Branched a-olefins
Ca-CIa cs-C8 cs-C8
d I-
-
'2
'
Figure 1.
!j
4' 5' 6'
I
'
WAVELENGTH , mi c ro n s
Distillate fraction, 292-93'
They have no other characteristic bands which could be seen a t these concentrations. There was no evidence of branched internal olefins (CRIRP = CHR3) a t 11.90 to 12.66 microns or &internal olefins a t 14.49 microns (3). Absorptivities of these bands are approximately one ninth and one sixth, respectively, of the absorptivity of the trans-internal olefin band at 10.36 microns (14). 2Methylalkanes were identified from the band a t 8.56 microns arising from the skeletal vibration of the isopropyl grouping (3). This assignment was supported by the fact that the absorbances of the 8.56-micron isopropyl
compds. examined 12
8 4
Wavelength, microns 11.02 (li uid) 11,02 (saution) 6 , 0 7 (liquid) 6.07 (solution) 8 56
10 36 11 27
4
n-Alkanes
Methylalkanes
a-Olefins
18.5 19.5 22.7 25.9 29.4 32.5 33 7 34 7 (38 3 ) 40 2 (43 7 ) (46 3 ) (49 0 )
5.1 6.3 8.2
10.5 13.2 15.4 16.8 19.5 22.1 24.3 27.9 31.5 (33.6) 36.2 (38,6) (41.2)
ANALYTICAL CHEMISTRY
700
Typical infrared spectrum obtained for a peak tube of 2-methylalkanes
absorptivity,* €
(126,5) 126,5 26.0 26.0 8 1 (113 9) (148 0)
olefins
a-olefins (7.9)
(31 5 ) (33.6) (36 2) (38 6)
a Values in parentheses were estimated from plots of molar absorptivity vs. carbon number. * The wavelength of this band for Cs is slightly less than 13.88 microns.
22 16
800
' " ' ~ 1 ! ! " 1 ! ' ' !
20-
Molar Absorptivities" at 13.88 Microns for Five Aliphatic Classes 2trans-Internal Branched
16.6 (17.7) (18 8 )
1
2- 40---
Molar absorptivities in liter mole-' cm.-* Values in parentheses were estimated as described in text. Table II.
'
-
Average Molar Absorptivities" for Diagnostic Infrared Bands Molar Carbon No. of
Aliphatic class a-Olefins
b
I
z
From a qualitative standpoint, the neutral oil aliphatic fractions were the same throughout the boiling range investigated. Figure 1 shows a typical infrared spectrum obtained. The qualitative information indicated that four classes of aliphatic hydrocarbons were present : e-olefins, trans-internal olefins, 2-methyl-a-olefins, and 2-methylalkanes (8). il close examination of the relative intensities of the 13.88-micron band in spectra of these four classes showed that in the neutral oil spectra there was some other substantial contribution a t this band. This was assigned to the fifth class of aliphatics, n-alkanes.
a
'
1500
60-
RESULTS AND DISCUSSION
Table 1.
-
' '
WA V E N U M6ER S , c m - 1 1200 1000 900
2000
3000
C.; CCD fraction collector tube 5
band and the 7.25-micron methyl band were greater for CCD tubes having maximum concentrations of saturates than for tubes having maximum concentrations of olefins. Further internal branching in the aliphatic neutral oil fractions was negated by the absence of bands in the 12.88- to 13.80-micron region (8, 11). Also, examination of pure compound spectra shows that extensive internal branching in a longchain aliphatic hydrocarbon cause? an increase in the intensity of several weak bands, and this was not observed in neutral oil aliphatic spectra (11). The validity and accuracy of this method for class analysis depend on the assumption that the terminal olefin bond, the internal olefin bond, the branched a-olefin bond, and the isopropyl grouping are on separate molecules. The infrared absorption bands selected for use in the quantitative analysis of the five aliphatic classes were 11.02, 10.36, 11.27, 8.56, and 13.88 microns. Their molar absorptivities are shown in Tables I and 11. RIolar absorptivities for the diagnostic bands remained constant with moiecular weight change. Their average values were determined from available samples, representing a wide molecular weight range. For the two classes, trans-internal olefins and branched a-olefins, there were no samples available above C8. From several literature sources, values for molar absorptivities a t their diagnostic bands were collected for these two classes and for e-olefins, for various molecular weight ranges. These data are shown in Table 111. In spite of the wide variation of actual molar absorptivity values, the ratio of the molar absorptivity of the diagnostic band of a-olefin to that of each of the other two olefin classes remained reaaonably constant. The average of these literature ratios was in good agreement with the values determined in the ])resent work for the limited number of compounds available in the C j to Cg range (1.13 for the ratio of (Y 'trans-in-
ternal olefins and 0.89 for a/branched a-olefins). Froin these ratios and the molar absorptivity at 11.02 microns for a-olefins, the molar absorptivities at 10.36 and 11.27 microns for trans-internal and branched a-olefins, respectively, were calculated. Iker's law deviations for the olefin hands are expected to be a t a minimum (Is), Anderson ( 2 ) observed that there is an increase in "functional group absorption coefficient" (where concentration is expressed as weight fraction of functional group) with molecular weight for the 11.02-micron band for Cr to Cs a-olefins. Above Cs t,his coefficient approaches a constant value. The present data for a-olefins confirmed this. Thus, the molar absorptivity value of 126.5 for the 11.02-micron band shown in Table I is applicable only above Cs. The other classes of olefins may also have this limitation and analysis below Cs can best be performed by using the absorptivities for the specific compounds occurring in the fraction to be analyzed. To determine the amount of nalkanes, the molar absorptivity a t 13.88 microns for each aliphatic class was needed. Unlike the diagnostic bands, the 13.88-micron band, assigned to the methylene rock in all aliphatics where there are four or more carbons in a linear chain, does not remain constant with molecular weight change. The molar absorptivity of this band for nalkanes increases linearly with increasing carbon number ($, IO). I n the present, work this relationship was plotted from the data for available n-alkanes and the plot used to determine the absorptivities a t 13.88 microns for compounds impure or missing in the CIS to Czorange. Very few samples of any class were available in this high molecular weight range and the purity of some was doubtful. Assuming the same linear relationship for the 13.88-micron band of a-olefins and 2-methylalkanes, plots of molar absorptivity a t 13.88 microns us. carbon number were prepared from the data for available samples. As for n-alkanes, reliable absorptivity values for the 13.88-micron band for these t'wo classes were obtained by extrapolation into the range of interest, Cl5 t'o Cz0. Because no trans-internal olefins or branched a-olefins were available above Cs and C7, respectively, the linear relationship between carbon number and molar absorptivity a t 13.88 microns could not be determined experimentally. Therefore, estimates of their contribut,ion a t 13.88 microns were made by assuming the trans-internal and branched a-olefin molar absorptivities a t 13.88 microns for each compound to be the same as the absorptivity of the a-olefin with the same number of CH2's in the hydrocarbon chain. Xn examination of ab-
Table 111.
Reference KO.
Literature Values of Molar Absorptivities" at Diagnostic Bands for Three Classes of Olefins 10 36 Ratios Xicron 11.27 trans11.02 Micron a/ branched a/trans- Branched internal Compd. range hlicron
investigated
c5-cs
cs-cu c5-c10
From ref. ( 1 )
C*
cs-clo c5-cs c5-cs c5-c11
cs-c17
Polgbut adienes Cr-Cs From ref. ( 1 )
cs-em0
a 0
c
133 139 133 133
~
100 147.2
a
0.86 0.76 1.17 0.89 0.94 0.87
0.84 0.86
0.87 __ 0.90
Liter mole-' cm.-1 Molar absorptivity in arbitrary units. Integrated values (liter mole-'
Table IV.
Boilin8 ranga, C. 275-78 282-RA
Quantitative Infrared Microanalysis of High-Boiling Aliphatic Neutral Oil Fractions
CCD tube NO.^ 5 FC
a-Olefins 21 18
transInternal olefins 9 10 7
a
a-olefins Internal 17.3b 0.89 63.7b 1.16 43.1b 1.41 1.08 1.33 137 1.17 159 1.12 1,12 160 1.05 1.38 5,460" 140 1.20 184.7 1.09 Av. 1 .17
olefins 16.7b 41.8* 35.6b 132-41 109 104
a-olefins 14.9; 48.6 50.3b 143-53 145 122 149 149 155 139 184 4 580" 120 160.4
Wt. 7% Branched a-olefins 1 2 2
%-Alkanes 36 51 3
2Methylalkanes 33 19
FC designates tube in fraction collector,
sorptivities at 13.88 microns for the lower molecular weight classes, CB, Gr and Cs, verifies this relationship. The complete data for the 13.88-micron band for all five classes of aliphatics, including both experimental data and values estimated from the linear plots, are summarized in Table 11. The Clj to Cz0 data were used for analysis of neutral oil samples. The CSto C14data served for calibration in this work but are also applicable to lower molecular weight mixtures. It is not known why values for molar absorptivities of 2methylalkanes at 13.88 microns are considerably lower than expected. There was no adequate test of the accuracy of this quantitative procedure by mixtures of known composition. It was not possible to test the method in the pertinent to Czo range because pure samples representing all five classes in this range were not available. I n fact, the only synthetic mixture that could be made uli representing all five claves was in the Cs to C8 range where the analysis is best performed by using the absorptivities for the specific compounds rather than the present procedure. However, in spite of these severe limitations, a variety of synthetic mix-
tures was prepared; these mixtures had fewer and fewer of the five classes as their carbon numbers increased. The maximum error observed for any class in any of these mixtures were about 470 (absolute). I n general, the errors were much smaller; for example, a mixture of 38.5% 2-methyldecaneJ 22.8% 1-undecene, and 38.7% undecane was determined to consist of 39.0% 2-methylalkanesJ 22.5% a-olefins, and 38.5yGn-alkanes. For each neutral oil distillate fraction a separate matrix was set up using values from Tables I and 11. Values for the diagnostic bands remained constant for each fraction. For the 13.88micron band, values were used for the compounds with boiling points closest to that of the neutral oil fraction being analyzed. Where the boilinq point fell between two compounds, their weighted average value was used. -4 few examples of the application of this procedure to the quantitative analysiq of aliphatics in low-temperature tar neutral oil fractions covering the boiling range 275" to 344' C.are presented in Table IV. Each distillate fraction was separated into aliphatics and aromatics by countercurrent distribution, and a VOL. 36, NO. 11, OCTOBER 1964
0
2217
separate infrared quantitative analysis was performed on the contents of each tube in the fraction collector. LITERATURE CITED
(1) American Petroleum Institute, Research Project 44, Carnegie Institute of Technology, Pittsburgh, Pa. ( 2 ) Anderson, J. A., Jr., Seyfried, W. D., ANAL.CHEM.20, 998 (1948). (3) Bellamy, L. J., “The Infra-red Spectra of Complex Molecules,” pp. 26, 34, 47, 50, Wiley, Sew York, 1958. (4) Binder, J. L., ANAL.CHEY.26, 1877 (1954). (5) Estep, P. A., Karr, C., Jr., Appl. Spectry. 16, 167 (1962).
(6) Friedel, R. A., U. S. Bureau of Mines, Pittsburgh, Pa., private communication. 1956. (7) Hampton, R. B., ANAL.CHEM.21, 923 (1949). (8) Harvey, SI. C., Ketley, A. D., J . Appl. Polymer Sci. 5 , 247 (1961). (9) Jones, R. N.. S.neclrochim. Acta 9. 235 (1957). (10) Jones, R. N., Sandorfy, C., in “Technique of Organic Chemistry,” A. Weissberger, Ed., Yol. IX, Chap. IV, Interscience, New York, 1956. (11) McMurry, H. L., Thornton, V., ANAL.CHEM.24, 318 (1952). (12) Richardson, W. S., J . Polymer Sci. 13, 229 (1954). (13) Richardson, W. S., Sacher, A., Ibid., 10, 353 (1953). I
(14) Saier, E. L., Cousins, L. K., Basila, M.R.ANAL.CHEM.35. 2219 (1963). (15) Saier, E. L., Pozefsky: A., Coggeshall, N. D., Ibzd., 26, 1258 (1954). (16) Silas, R. S., Yates, J., Thornton, V., Ibid., 31, 529 (1959). PATRICIA A. ESTEP CLARENCE KARR,JR.
.
Morgantown Coal Research Center U. S. Bureau of Mines Department of the Interior Morgantown, W. Va. Division of Analytical Chemistry, 147th Meeting, ACS, Philadelphia, Pa., April 1964.
Complete Gas Chromatographic Analysis of Fixed Gases with One Detector Using Argon as Gas Carrier Addendum SIR: Since the publication of my article ( 2 ) previous articles by Hernandez, Strand, and Vosti, (1) and by Murakami (3) have come to my attention. I regret that these two articles were not included in my original bibliography. Both Vosti, e t al. and Murakami use two detectors, each equipped with a single filament. I n the analysis of fixed
gases the filaments of these two cells serve as opposing variable resistances in a bridge circuit. One cell is placed at the exit of the silica gel column and the second cell after the molecular sieve column. As a gas component is detected by either filament the recorder deflection is maintained in the same direction by a polarity reversing switch.
LITERATURE CITED
K.H., Strand, J. B., Vosti, D. C., Food Technology 15, 29 (1961). (2) Manka, D. P., ANAL. CHEY.36, 480 (1964). (3) Murakami, Y., Bull. Chem. SOC. Japan 32, 316 (1959); C. A . 54, 3045d (1960). DANP. MAKKA Graham Research Laboratory Jones & Laughlin Steel Corp. Pittsburgh 30, Pa.
(1) Hernandez,
Time Dependence of A.C. Polarographic Currents An Exchange
of Comments
SIR: Hung and Smith (6) have reported preliminary theoretical and experimental results concerned with the dependence of a.c. polarographic currents on drop time. We wish to point out that we have presented (1) a numerical analysis of Matsuda’s equation for the a x . polarographic wave ( 8 ) , and have indicated that under quasireversible conditions the current at the summit potential is drop-time dependent. Senda (IO) has discussed the time dependence of a.c. polarographic currents and has suggested its use to distinguish between quasi-reversible and reversible electron transfer reactions. We have shown also that this drop-time dependence for the quasireversible case may be predicted from a treatment based upon that of Breyer and Bauer ( 2 ) . Electrode-surface concentrations- of oxidant and reductant were derived using a quasi-steady-state approach (4, 12) and the results for the 22 18
ANALYTICAL CHEMISTRY
a.c. polarographic wave compare favorably with those of Matsuda ( 8 ) . Also, we have derived an equation for the a.c. polarographic wave in the case where a fast, irreversible, monomolecular, chemical reaction follows a reversible electron transfer reaction. The current decreased with increasing drop-time (1). Using this equation, we have measured the rate constant for the formation of the cadmium-EDTA complex by oxidation of cadmium amalgam in EDTA solution. Our results compare well with that of Koryta and ZAbransk$ (7) who used the shift in d.c. polarographic half-step potential to obtain their result. The rate constant for the substitution of calcium ions in the europium(I1)EDTA complex, formed by reduction of the europium(II1)-EDTA complex, has also been measured by a x . and d.c. polarography, The two methods do not
give the same result, and the phase angle measurements indicate that a double-layer effect is responsible for the discrepancy. One of us, with Florence (5j, has reported drop-time dependence of the a x . polarographic current for the reduction of 5-sulfo-2-hydroxybenzeneazo-2-naphthol. It was suggested that the mechanism of the reduction involved disproportionation of the hydrazo species. We have observed numerous other cases of drop-time dependent, a x . polarographic currents. In all of these, the electrode-surface concentration of the oxidant or the reductant is subject to a mass transfer process other than, or combined with, diffusion. The references quoted below refer to the papers in whichthis has been established. For example, the a.c. polarographic wave height for the reduction of aquo cadmium ions formed by dissociation of