228
ANALYTICAL CHEMISTRY
It appears that the conjugated triene correction term is effective up to about 6% conjugated triene acids in the presence of about 20% polymerized styrene. The error in the amount of polymerized styrene increases as the differences between the terms of Equation 5 become of the same order as the photometric error and as the contribution of the trace amounts of conjugated tetraene becomes appreciable; the absorptivity of the conjugated tetraene acid (2, 3) is lo4 greater than that of polystyrene a t 291 mp.
Table 111.
filtrate is concentrated, transferred to a separatory funnel, and extracted with ether. The ether layer, which contains polystyrene, fatty acids, and interaction products of these, is evaporated to dryness. A sample of this fraction is weighed and dissolved in cyclohexane for spectrophotometric examination. ACKNOWLEDGMENT
The authors wish to acknowledge the assistance of William G. Deichert with sample preparation and separation and of Mrs. E. S.Davis and John G. Koren with the spectrophotometric work.
Effect of Conjugated Acids
Per Cent Conjugated Sample Triene Tetraene Description acids acids Linseed 0.12 0.040 19 : 1, linseed-tung 2.3 0,05! 6:1,linseed-tung 6.2 0.0711 3:1,linseed-tunc: 1 1 . 8 0.103
Per Cent P o l y s t y r e n L BY By ultrar.iolet preparation sp~ctiopliotornetry 25.0 23. 1 21.3 23.3 2 2 . ii 19.3 20.2 13 7
APPLICATION TO ALKYD RESINS
The details of the application of this spectrophotomrtric method to the analysis of resins have been described bj- Stafford and coworkers (10) and are summarized here. The ,\ST111 method (1)is applied to the alhvd resin as received. The sample is saponified and the dibasic acids are isolated hv prwipitation a9 insoluble potassium salts, in Lvhich the phthalic acid ma! 1)c determined by the ultraviolet spectrop3otometric Shreve method (9). The filtrate is neutralized with concentrated hydrochloric acid, and the precipitated potassium rhloride is filtered oft. The
LITERATURE CITED
( I ) Ani. SOC.Testing Materials, Method D 563. (2) Bradley, T. F., and Richardson, D., Ind. Eng. Chem., 34, 237 (1942). (3) Brice, B. -4.,and Swain, AI, L., J . Opt. SOC.Amer., 35, 53% (1945). (4) Brice, B. A., and Swain. 31. L., Oil & Soap, 22, 219 (1945). (5) Hirt, R. C., and King, F. T., Ax.ir.. CHEM.,24, 1545 (1952). (6) Kappelmeier, C. P. A , and Van der Seut, J. H., Chem. Weekblad, 47, 157 (1951). (7) JIcGovern, J. J.. Grim. .J. AI., and Teach, W. C., AN.AI..CHEM., 20,312 (1948). (8) Yewell, J. E.,Ibid., 23, 445 (1951). (9) Shreve, 0. D., and Heether, AI. R., Ibid., 23, 441 (1951). (10) Stafford, R. TI-,, Hirt, R. C., and Deichert, W.G., Division of Paint, Plastics, and Printing Ink Chemistry, 126th Meeting, .‘hERIC.AN CHEMICAL SOCIETY, New York, 1954. (11) Swann, 31. H.. . ~ x . A L . CHEM.,25, 1735 (1953). (12) Witschonke, C. It., Ibid., 24, 350 (1952); 26, 562 (1954). R E C E I V E for D review August 25, 1951. Accepted November 9, 1954. Presented before the Pittsburgh Conference on .4nalytical Chemistry and h p plied Spectroscopy, March 5 , 1951.
Far Ultraviolet Absorption Spectra of Aromatic Hydrocarbons L.
C. JONES, JR., and
L. W. TAYLOR
Wood River Research Laboratory, Shell O i l Co., Wood River, 111. A survey of the anal) tical potentialities of absorption spectroscopy in the far ultraviolet region has been initiated. In the first phase of this program the far ultraviolet (1700 to 2300 A.) spectra of 69 pure hydrocarbons have been measured either in the vapor or solution state with a recording vacuum spectrometer, employing a 1-meter concave grating and a photoelectric detector. Experimental methods are described and spectra of representative paraffins, naphthenes, cyclic and noncyclic mono-olefins, allenes, conjugated and nonconjugated diolefins, acetylenes, benzenes, naphthalenes, and polycyclic catacondensed aromatics are presented. Useful correlations between the spectra and structure of these hydrocarbons are discussed. It is concluded that far ultraviolet absorption spectroscopy is a promising new tool for the analyst.
A
BSORPTION spectroscopy in the near ultraviolet (2100 to 4000 A,) is one of the most widely used physical methods of
analysis. Its popularity can probably be ascribed to two factors: the large number of organic compounds with characteristic absorption in this region, and the availability of well-designed, reliable, and relatively inexpensive commercial instruments. The value of the method in the analysis of petroleum fractions is enhanced by the existence of extensive compilations of accurately measured and uniformly presented spectra of pure hydrocarbons.
This’is in marked contrast to the present status of absorption spectroscopy in the far or vacuum ultraviolet region. Despite a half century of academic research in far ultraviolet spectroscopy, the technique has not, t o the authors’ knowledge, been applied by the analyst. The lack of adequate instrumentation has undoubtedly been a large factor in delaying such application. Until very recently intensity measurements in this region have depended upon photometry with special photographic emulsions of very poor reproducibility. Consepently, emphasis in the early research was largely directed toward accurate meamrement of absorption band positions and upon calculation of molecular energy level*. The intensity data obtained in these studies were not of the degree of accuracy required in analytical applications Recent advances in photographic photometry in the far ultraviolet have improved this situation considerably, but the fact remains that most of the published spectra are of limited value to the analyst. The qualitative and semiquantitative features of these spectra suggested, however, that the far ultraviolet region had potential utility in the analysis of hydrocarbons. The photographic measurements of Carr and Pickett and their collaborators at Mt. Holyoke College (1, 12, 19, 35) on the aliphatic olefins, diolefins, and cyclic unsaturates, and of Platt, Klevens, and others a t the University of Chicago on the aromatics (16, i7, 21, 23) and other unsaturated hydrocarbons (14, 18, 24, 26) are of special interest in this connection. Measurements of analytical accuracy in the far ultraviolet
V O L U M E 27, NO. 2, F E B R U A R Y 1 9 5 5 have been made possible in the past few years by the development of recording photoelectric spectrometers ( 4 , 10, 11, 13, 15, 34, 37-39). With such an instrument a study has been undertaken of the analytical possibilities of absorption spectroscopy in the 1700- to 2100-A. region. The lower wave-length limit has been determined both by the transmittance limits of available solvents (22, 2 7 ) and by the lower limit of the hydrogen continuum. The measurements have been extended to 2300 A. on the long wave-length side to permit comparison with previously available data from near ultraviolet spectrometers. Since the primary interest was in the field of petroleum chemistry the first phase of this program has involved the measurement of the spectra of a large number of unsaturated hydrocarbons. The spectra of 69 representative hydrocarbons are reported in this paper. Although this phase of the program is incomplete the results are believed of sufficient interest to warrant a preliminary report a t this time, EXPERIMENTAL METHODS
Spectrometer. A schematic diagram of the spectrometer is shown in Figure 1. The basic component is a Baird vacuum ultraviolet monochromator. A detailed description of this instrument has been given by Tousey et al. (37). Briefly, it consists of an evacuated 1-meter concave grating spectrometer. The grating is an original Johns Hopkins ruling with 15,000 lines per inch and a ruled area of 1 1 / 2 X 3 3 / 4 inches. The reciprocal dispersion is 16.9 A. per mm. Wave-length scanning is accom-
CONTROLLER
7f' -L
. 1 S F
LiF LENS
w
TO GAS TO PUMPS HANDLING SYSTEM
iY;F
Figure 1.
Schematic diagram of far ultraviolet spectrometer
MERCURY LIQUIDS SINTERED DISK --c
TO TRAPS AND PUMPS
SAMPLE STDRAOE
il
1
---8. FOR LOW BOILING LIQUIDS { ABSORPTION
m W
? +
1 1
To MERCURY MANOMETER
TO TRAPS *DIFFUSION AND MECHANICAL WMPS
Figure 2.
Gas-handling systems
229 plished by translation of the grating about ,the circumference of the Rowland circle. The wave-length drive permits the spectrum to be scanned a t either 36 or 100 A. per minute over the 1700- to 4100-A. region. The monochromator has a counter which reads very nearly in Angstroms. If an extensive spectral region is to be covered, however, it is necessary to apply small corrections to the counter readings because of the slight nonlinearity of dispersion of the grating. A correction curve was constructed from measurements of the counter readings for the central beam and for near ultraviolet lines of the mercury arc. These data were fitted by the method of least squares to a power series form of the equation for the dispersion of a grating a t normal incidence: A =a
+ bR + cR3
(1)
where X is wave length, R is the counter reading, and a, b, and c are constants. The calibration equation obtained in this manner has been used to calculate the wave lengths of the SchumanRunge bands of oxygen and of the 1600-A. series of hydrogen lines. Agreement with published results was always within 1 .4., which is the accuracy claimed for the instrument by the manufacturer. The radiation source is a windowless hydrogen discharge tube, similar in many respects to those described by Powell (28) and by Johnson, Watanabe, and Tousey (16). However, where they used direct current excitation of the arc, an 1800-volt, 60-cycle power supply was used here to obtain 120 pulses of light per second. This permits use of a stable alternating current amplifier in connection with the photomultiplier detector to obtain improved discrimination against the thermal emission of the multiplier tube and a better signal-to-noise ratio. The lamp is mounted on a bracket which may be rotated about the entrance slit of the spectrometer. This is desirable, since in order to reduce stray light t o a minimum the lamp is fitted with a baffle to restrict the cone of illumination entering the spectrometer to that accepted by the grating a t any one wave length. In scanning a wide wavelength region the grating will move out of this cone, so that it may be necessary to readjust the position of the lamp. Such adjustment is not necessary if the measurements are restricted t o the 1700- to 2300-.4. interval as in the present case. The intensity of the lamp is monitored by an ultraviolet-sensitive phototube and maintained constant within about 1% by a servo circuit described by Pondrum and Robertson (26). Located immediately behind the exit slit is the cell compartment which accommodates kinematically mounted absorption cells up to 6 cm. in length for liquids and gases. These are of the lead spacer, amalgam seal type commonly used in infrared spectroscopy, except that they are fitted with windows of synthetic lithium fluoride. Convenient cell lengths for solutions are 0.1 and 0.5 mm. The exact lengths of these cells were determined by measurement of the absorbance a t 2500 A. of the cells filled with a solution of diphenyl. An aliquot of the diphenyl solution was then diluted quantitatively so that its absorbance a t the same wave length might be measured in the 10.0-mm. Beckman cells. The cell thickness could then be calculated from Beer's law. All of these measurements were made with a Beckman DU quartz spectcophotometer. .It has previously been established that this method gives thicknesses in agreement with the infrared interference fringe method (36) within the precision of the latter method. The vapor cell used in the present work is 30.8 mm. in length. I t is connected to either of the external gas-handling systems shown in Figure 2 via spherical joints brazed into the cell compartment wall. A lithium fluoride double convex lens images the exit slit on the photocathode of the photomultiplier. The latter is a special fused-silica envelope 1P28 tube (RCA Development Type C-7139) identical with those described by Dunkelman and Lock (10). The multiplier has excellent sensitivity to 1540 A., where i t cuts off abruptly. It is operated a t a total dynode potential of 840 volts. The output of the multiplier is amplified by a con-
ANALYTICAL CHEMISTRY
230
*
Figure 3.
Typical recorded spectrum
ventional, aharply tuned, 120-cycle amplifier supplied by Baird Associates and modified so that the gain might be changed reproducibly and stepwise by factors of & The output of the amplifier is rectified and fed into a Brown strip chart recorder. Vapor-Handling Technique. Vapors of condensable compounds are handled in the gas-handling systems of Figure 2. Known amounts of samples which may conveniently be handled as liquids are introduced into the evacuated absorption cell and associated large bulb through a mercury-covered sintered-glass disk (Figure 2,A) by means of calibrated micropipets of 5.0- to 0.1-pl. size. The larger pipets were obtained from Microchemical Specialties Co. and the manufacturer’s calibration was accepted. The smaller sizes were made especially for the present application and were calibrated gravimetrically with mercury and the calibration was confirmed by means of a mass spectrometer. The more volatile compounds (C, and lighter) are handled in the system shown in Figure 2 3 . The sample is vaporized into the evacuated manifold a t A or B to a pressure of 30 mm. or more. Noncondensable contaminants such as air are removed bv freezing the sample in the trap, pumping off uncondensed gases, and re-evaporating. This process is repeated until no residual pressure is detected on freezing out the sample. The sample is then evaporated once more and the pressure measured. A known volume of the vapor is then isolated a t this pressure in bulb C and the remainder of the system evacuated. The vapor is then expanded from C into the large bulb and the absorption cell. If necessary, a second such expansion may be made. In this manner static pressures as low as 0.05 mm. may be obtained with an accuracy of about 1%. The gas-handling system may be heated electrically to facilitate out-gassing. The effect of absorption-desorption phenomena is minimized by carrying out most of the spectral measurements a t pressures below 2 mm. of mercury. Satisfactory performance of this method of gas handling has been confirmed by obtaining linear plots of absorb ance us. calculated pressures for several olefins a t a number of wave lengthe. Good agreement between the two methods of gas handling was obtained nhen both were used with the same sample. Preparation of Solvent and Solutions. %-Heptane is a satisfactory solvent for use in this region of the spectrum (26). -4STM knock test reference fuel ( S ) , n-heptane (Special Produrts Division, Phillips Petroleum Co., Bartlesville, Okla. ), is purified by percolation through Davison 923 grade silica gel. One pound of gel is required per pint of heptane. Solvent prepared in this manner has good transmittance to 1715 A. in 0.1-mm. thickness and to 1740 A. in 0.5-mm. thickness. Solutions were prepared by volumetric dilution of the sample with the solvent. Ultramicropipets and 5.0-ml. volumetric flasks were used for this purpose. Since these pipets are calibrated “to hold” rather than “to deliver” they were carefully rinsed several times with thtx solvent and the rinsings added to the volumetric flask. Solid
samples were weighed into the volumetric flasks with a microbalance. Operating Technique. All spectra were scanned a t 36 A. per minute. A blank energy ( l o ) run was made with an evacuated sample cell in the case of the vapor spectra or with a cell filled A H30. VPPOR PHASE SPECTRA SPECTRAL SLIT
A
Z G 10,cQoI
I
-I
1
I
\I
’
F. I-HEXENE
c~c-c€€-c
I
G. 3-METHYLL-PEMENE
I 10 I-PENTENE
I
I
h
\
c
DIMETHYL”?-BUTENE
Figure 4. Far ultraviolet spectra of vapors of ethylene and C3 t o Cg monoalkylethylenes
231
V O L U M E 2 7 , NO. 2, F E B R U A R Y 1 9 5 5
,
with the solvent in the case of solutions. The sample vapor or solution was then introduced, the region rescanned, and a second energy (I) trace made. T o ensure detection of errors due to unexpected changes in intensity of the lamp, the energy was also measured in a region where the sample was known not to absorb appreciably immediately before each sample run. A constant spectral band width of 1.7 A. (slit width = 0.1 mm.) has been maintained in all work with vapors. The spectral band width was 5.1 A. (slit width = 0.3 mm.) for solutions. The larger slit width was chosen in the latter case in part because of the poorer transparency of the lithium fluoride windows used in the liquid cell and in part because of the improved signal-tonoise ratio obtained in this manner. S o n e of the solution spectra examined to date has shown any structure which would justify use of narrower slits. A typical vapor spectrum recording is shown in Figure 3. The gain of the amplifier was changed in the middle of the record to maintain a reasonable signal strength
Table I.
a t the shorter wave lengths. The abrupt cutoff of the photomultiplier envelope a t 1540 A. can be seen. The zero line shown was obtained by scanning with a metal shutter in the light path. The near-coincidence of this line with that of the sample spectrum a t 2100 A. and with the blank spectrum below the cutoff is evidence that scattered radiation of unwanted wave lengths is of negligible intensity.
I
1
1
I
1
VAPOR SPECTRA
1
Source and Purity of Standard Samples
Compound Ethylene Propene 1-But.ene &&Butene trans-2-Butene % M e t hylpropene I-Pentene cis-2-Pentene trans-2-Pentene 2-Methyl-1-butene 3-Methyl-I-butene 2-Methyl-2-butene 1-Hesene cis-2-Hexene trans-2-Hexene cis-3-Hexene trans-3-Hesene 2-hlethyl-1-pentene 3-Met hyl-1-pentene 4-Methyl-1-pentene 2-Methyl-2-pentene 3-I\Iethyl-cis-2-pentene 3-Methyl-trans-2-pentene 4-hIethyl-cis-2-pentene 4-Methyl-trans-2-pentene 2-Ethyl-1-butene 2,3-Dimet hyl-I-butene 3,3-Diniethyl-l-butene 2,3-Dimethyl-2-butene Cyclopentene Cyclohexene 1,3-Butadiene 1-cis-3-Pentadiene 1-trans-3-Pentadiene 2-Methyl-1 3-butadiene 2,3-Diniethyl-l,3-butadiene 1,2-Butadiene 12-Pentadiene 2,3-Pentadiene 1,+Pentadiene 1,5-Hexadiene I-Butyne Benzene Toluene o-Xylene m-Xvlene p-xylene 1 -Jlethvl-2-ethvlbenzene
Source Phillips" Phillipsa Phillipsa Phillipsa Phillipsa Phillipsa .4PI 281 b .%PI 2826 .%PI 283b .%PI284b .%PI285b .%PI 2866 .APT 319b ..\PI 5 2 6 b .-\PI 5276 .\PI 528b ;\PI 329h .-\PI ;?Oh APT 5x1 h
APT 312b APT X . 7 b A P I 534h
.%PI .i.75b A P I 3.776 .-\PI 3iRfih .-\PI 3'386 ..\PI .i39h API 2876 API X O b API 28Rh API 3 2 2 b Shell Oil .$PI ,5636
..\PI 3 x 5
.iPT %4Ob .%PI .i71)b .iPI 5126 .IPl ;69b .\PI 5 5 8 h ..\PI . i l X b ..\PI 3 5 3 h ..\PI 5 l 4 b .\PI 2 1 O h .iPT 21 1 b ..\PI 213h ..\PI 214b .%PI 213h . i P I L'4lih ..\PI 2471, API 248h API 2-19 API 2 3 O b .4Pr 7.i1 h ..\PI j R C , h .%PI 573i' .-\PI 5 8 , j h API .j86b API S37h .iPl n77h APT d78b API .i79b f'
1 2 3 5-Tetramethylbenzene 1:2:4:5-Tetrameth~lbenzene n-Deovlbenzene T etrah Tetrafin Kaphthalene I-Methylnaphthalene 2-Methylnaphthalene 1,2'-Dinaphthyl
V . S. Bur.
Phenanthrene
11ines c This laboratory
Anthracene
This laboratory
Chrysene
U . S. B u r .
1,Z-Benzanthracene
Mines c Eastman Kodak
20-Methylcholanthrene
Eastinan Kodak
2,2,4-Trimethylpentane n-Decylcyclohexane
co.
Purity, Mole % 99.96 99.7 99.9 98.9 96.2 99.4 99.74 99.89 99.91 99.86 99.76 99.94 99.86 99.80 99.83 99.87 99.94 99.81 99 71) 99 82 99.91 99.85 99.86 99.92 99.75 99.90 99.86 99 91 99.9n 99.97 99.98 99 99.92 99,92 99.96 99.94 99.92 99. 6li 99 , 8 5 99.93 99.84
99.87 99.98 99.97 99.99 99.93 99 .9R 99.73 99.57 99.87 99,982 99.67 99.96 99.89 99.92 99.8ii 99.80 99.86 99.96 99.92 99.91 1f.p.. 76.477.4= h I . p . , 100.9101.20 1z.p.. 216 0216.2' C. hl.p., 25-1256' C. A L P . , 159160° C. ?rI.p., 176178' C . 99.96 99.8ti
c. c.
Figure 5 .
Far ultraviolet spectra of vapors of Cn to CS 1,2-dialkylethylenes
Molar absorptivities, E , were calculated from the recorder pen deflections by the usual relationship: E = (log,, Io/I)/bC (2) where b is the cell length in centimeters and c is the concentration in moles per liter. No correction was required for absorption by the solvent, as the solutions were dilute and the transparency of the solvent was high. Due allowance was made for temperature effects and nonideal behavior of the gases in computing the concentrations of the vapor samples. I n each case the spectra reported represent the average of three consistent runs, usually made on different dags as an additional safeguard against fortuitous repetition of error. Molar absorptivities obtained in this manner normally have a standard deviation of about 375, so that the 95% confidence limits for the average of the three runs should be approximately *5% unless some unexpected source of systematic error exists. Photodecomposition of the sample due to prolonged exposure to high-intensity far ultraviolet radiation was one of the major problems of early workers in this field. This problem has been eliminated completely by the present arrangement of having the absorption cell beyond the exit slit of the spectrometer where it is exposed only to the low-intensity dispersed radiation. The spectra of many of the samples have been scanned repeatedly without finding any detectable change in the absorption spectrum resulting from irradiation. Standard Samples. The source, and when known, the purity of the compounds used in this work are shown in Table I. Exact data on the purity of some of the samples were not available. I n these cases the melting point range as measured with a Iipfler micr? hot stage (Arthur H. Thomas Co.) and polarizing microscope is given.
ANALYTICAL CHEMISTRY
232 COhlPARISON OF PRESENT WORK WITH PUBLISHED SPECTRA
tained in this work for the Ca and Cj olefine and benzene vapoi are usually lower than those of the N t . Holyoke workers (19). The cause of this discrepancy is obscure, but may well lie i l l Thirty-two of the 69 compounds whose spectra are reported differences in sample-handling techniques. This is suggested b v here have been examined by other investigat,ors. The results are the fact that agreement seems to be best where the sample preecompared with those given in the literature in Table 11. Of sures were highest (of the order of 0.1 to 3 mm.) and pooiest at these, Carr and Pickett’s data on the C,, C6, and CS olefins the lowest pressures. A11 of the measurements were made a t (1,12,56) and Zelikoff and Watanabe’s spect’rumof ethylene ( 4 0 ) pressures about ten times higher than those of Carr and of Pickett have been reported since completion of the present study. as a result of a tenfold difference in the length of the cells. T h r I n general, it has been found that agreement between wave absorption bands for many of these compounde are so broad and lengths of sharp absorption bands (such as are found in the spectra flat that determination of the wave length of maximum abeorpof ethylene and 1,3-butadiene) found in the present work and tion is often subject to considerable uncertainty. Occasional those published earlier has been excellent, usually within 1 -4. discrepancies of the order of 10.1. aie thu- not surprising. Agreement of band intensities has been somewhat less satisS o consistent trend can be found in comparing the solution factory. The only compound of the present series whose specspectra 15ith those of Platt arid his colleagues. Differences 111 intrum has been measured by photoelectric t,echniques is ethylene. tensity are random and average about 15%. This is not inconThe present intensit,y data are systematically about 129;b higher qistent with the accuracy claimed for his intensity measurements. than those published earlier although the wave lengths agree Again nave lengths of absorption bands agree as \Tell as might be n3hin 1 -4.in all cases. On the other hand, peak intenshie$ ohexpected with two notable exceptions. Jacobs’ and Platt’s s p e c t r u ni f o r 2-methy1-1,3Table IT. Comparison of Present Work with Published Spectra hutadiene is shifted about 20 V a v e Length, A.5 Molar Absorptivity -1. to shorter wave lengths This This Spread, compared to that of Figure 9, work Lit. Coiiipound State work Lit. % while their spectrum for I-czsV 1704 17,100 14,800 f14.4 1703 Ethylene 11,200 10,400 - 7.4 1730 1730 3-pentadiene is shifted about 1744 1743 16,800 13,600 721.0 1,120 900 +21.7 1778 1778 10 A. in the opposite direc1802 490 435 111.9 1802 tion. 1826 260 230 +12.2 1825 1853 1882
1854 1881
99 89.5
10fi 38
A v. 1-Butene cas-2-Butene trans-2-Butene Z-.\lethylpropene 1-Pentene cas-2-Pentene t i ans-2-Pentene 2-Methyl-1-butene 3-Methyl-1-butene 2-Methyl-2-butene 1-Hex en e EiZ-Hexene 2.3-Dimethyl- 1-butene 2.3-Dimethyl-2-butene Cyclopentene C yclohexene 1.%Butadiene
l-iralls-3-Pentadiene I-cts-3-Pentadiene 2-l\Iethyl-1,3-hutadiene Benzene
v v
1’ V
v
V V
v v v
\‘
v
V
v V v v
v v v v
v v
V
174,5 1740 1780 1882 1780 1773 1812 1883 1753 1775 1772 1790 1883 1870 1828 1755 2158 2094 2035 1977 1864 1754 1803 2148 1895 2105 1910 1892 1760
1752 1754 1778 1883 1767 1767 1802 1879 1761 1768 1779 1787 1876 1865 1831 1754 2158.7 2049.1 2035.2 1977.8 1863.9 1754.7 1807 2138 1883 2113 1890 1871 215.5 1769
27,500 11,000 19 ,900 t i , 400 7,600 1‘). 800 08,000
17G4
17DG
ii7, ,500
1791
1790
109,000
215.5
11.200 1li,000 13,000 11,300 13,700 13.500 12,000 9.700 14,000 11,000 10,800 14,700 9,700 13,!i00 18, ti00
8,000
..
...
15,300 21.900 14,100 14,800 15,850 16,400 13,300 11,200 13,200 10,400 10,720 15.140 10,960 11,220 9,340 8.310
-32.2 -31.1 - 8.1 -26 8 -12.0 -19.4 -11.8 -14.3 7 5.9 A 5.6 0.7 - 2.9 -12.2 +32.6 f37.1 - 3.8
... ...
... ... ... ...
..
..
..
A ;000
r, :io0
1808 70,000 Sb 1840 1835 00,000 2042 2045 8,000 1885 s .iJ ,000 1890 Toluene 2050 7.800 2047 S 1905 3,000 1918 n-Xvlme 1935 S g1,000 1930 LIZylene 1925 J 1920 57,000 p-Xylene S 5 0 , 000 1990 2005 1,3,5-Triniethylbenzene 184Om s 1840m .5 , ti00 Naphthalene 2210 2202 122,000 S 30,000 1872 1875 Phenanthrene 14,000 1965m 1970m 2122 37,000 2115 S 1861 21.500 1885 .4nthracene 1960m 1975m 4,400 2203 2208 10,000 S 1831 1848 49,000 C hrysene 1940 1948 18,000 12.400 2030m 2030m 22,500 s 1822 1825 1.2-Benzanthracene l90Om 18,000 1910m 27,000 2028 2022 a Wave lengths correspond to major absorption band peaks, except when between absorption bands. b Solution in n-heptane. 18’14
++ 3fi.7. 8 +12.4
2.5,400 7,400 20,100 R,200 !i ,000 21,000 70,000 91,000 77,600 96,000 12R.000 127,000
m ,oon
47,000 li ,900 5.5 , 000 8,200 59,000 76,000 r,2.000
4.5 ,000 ,j, 000 133,000 31,000 12.500
+
...
+i. .58... 01 f39.1 - 1.0 i3 . 2 i23.5
- 5.R
-11.1 -29.0 -13.9 -35.0 -13.2 -13.3 1.4 L24.3 A14 8 0 0 - 5.0 -10.7 -49.0 - 8.4
+
121.4
L11.1 - 8.6 - 3.3 f11.3
33.000 +11.4 32,000 -39.2 - 27. .5 3,800 14,500 -3R.7 .50,000 - 2.0 18,300 - 2.7 12,000 t 3.3 23,000 - 2.2 1R.000 +32.3 21.000 +25.0 followed by m, indicating n ~ i n i m u n ~
CORRELATIONS BETWEEX F9R ULTRAFIOLET SPECTRA AND MOLECULAR STRUCTURE
The far ultraviolet spect.r:i of related groups of unsaturated h y d r o c a r b o n s s h o w regularities which should be useful b o t h i n q u a l i t a t i v e analysis and in quantitative estimat,ion of these compound^ in complex mixt,ures. Some of the more striking correlations will be d i s c u s s e d i n detail. Noncyclic M ono-olef i n s. The vapor phase spectra of d l the aliphatic mono-olefins from ethylene through the hesenee have been measured (Figures 4 to 6). Sis of the hexeiies have also been examined i n nheptane solution (Figure i ) . IYith t,he exception of ethylene, which is a special case, the olefins have remarkably similar spectra. Each has a broad, intense, and flat absorption band ( e about 10,000) with maximum between 12.3 and 1940 -4. According to current theories this band result,s from a transition involving one of the nonlocalized unsaturation or rr-electrons in m o l e c u l a r o r b i t a l s . I t is generally designated by the notation S-V. Overlaying the S+V band is it series of moderately intense v i h r o n i c
V O L U M E 2 7 , NO. 2, F E B R U A R Y 1 9 5 5
233
Table 111. Average Position and Peak Intensity of Absorption Bands of Vapor Spectra of the Alkenes Olefin Type 1-hlkenes c Is-Z-Alkenes trans-2-Blkenes 2-.llkyl- I-alkenes
A
A x . . Position of M a x . , A.
Molar Absorptivity. Liters/Mole C m . 1 1 , 8 0 0 zt 1200 1 2 , 3 0 0 5 200 1 1 , i O O i 800 8,900 1200
+
1730 23 1i1G 5 13 1790 I 10 1873 5 1 3
loo.
+
1
-
A 1O.OOO~
1
VAPOR SPECTRA SPECTRAL SLIT WIDTH 17A
1
A
(Figure 6). Only one of the tetraalkylethylenes was included in the present study. I n preparing Table I11 the average peak position and intensity have been determined from smoothed curves,-i.e., the vibronic (or Rydberg) structure of the bands has been averaged out to produce a curve with a single maximum. This procedure seems justified, as the principal uses of such correlations are in confirmation of structure of new compounds or in the estimation of total concentration of a given type of compound in a complex mixture. Since the vibronic structure of the spectra varies appreciably within a given series, it tends to be averaged out in complex mistures. On the other hand, it cannot be predicted either from empirical or theoretical considerations with enough certainty to be of value in the determination or confirmation of the structure of a new compound. However, the vibronic structure is more pronounred for compounds containing two or more alkyl substituents (Figures 5 and 6 ) on the ethylenic carbons than for the monoalkvlethylenes (Figure 4). Within a given series there seems to he some tendency for the peak intensity to increase with increasing symmetry of the molecule. Thus it increases in the series: 4-methyl-trans-2-pentene < trans-2-hexene = tranB-2pentene < trans-2-butene = truns-3-he1ene. Another effect has been pointed out by Carr and Stucklen (8, 9): a nearly linear relationship evists between the frequency of the absorption band having the longest wave length and the number of alkyl substituents on the ethylenic carbons. This bathorhroniic shift is greater for the cis than for the trans configuration. A 100
SOLUTKN SPECTRA SOLVENT: N-HEPTANE SPECTRAL SLIT WIDTH 5.1A
1
A 0,oOo7
Figure 6. Far ultraviolet spectra of vapors of Cd to CS 1,l-dialkylethylenes, trialkylethylenes, and
tetramethy lethylene
and Ilydberg series bands ( 5 , 29-32). These are very sharp in tlie case of ethylene but diffuse for the substituted ethylenes. Ftir analytical purposes it is convenient to group the olefins according to t,he number of alkyl groups att,ached to tlie carbons of the double bond. Some of these groups must also be subdivided into cis and trans subclasses. \Tithin these groups t,he far ultraviolet spectra are similar bot,h qualitatively and quantitativel>-,especially if the lowest member of each series is excluded. Thew similarities are immediately apparent upon examination of Figures 4 to 6. Kithin a given class, e.g., 1-alkene or ris-2alkene, all of the spectra lie within an area representing an intensity spread of about 10 to 20% of the mean. The peak intensities fall within an even narrower range as shown in Table 111. The spectra of the tri- and tetraalkylethylenes are not included in the correlations of Table 111, as an insufficient number of these compounds has been run t o establish reliable trends. Thus t,he spectra of three of the four t~rialkylethylenesshow a strong family ri~semblancewhile the fourth, 2-methyl-2-pentene, is anomalous
WAVELENGTH IN ANGSTROMS
Figure i.Far ultraviolet spectra of n-heptane solutions of representative hexenes
The spectra of n-heptane solutions of six of the hevenes are given in Figure 7 . I n going from vapor phase to solution spectra the peak intensity decreases slightly, a compensating increase in band 15 idth occurs and the bands shift about 40 d.to longer w a w lengths. The integrated intensities of the bands remain virtually unchanged. Thus for 1-heuene the oscillator strengths, f ,
f
=
4.32 X 10-9 f E Y dv
(3)
are computed to be 0.285 and 0.275 for vapor and solution spectra, respectively. I n Equation 3 Y is the frequency in cm.-l It is obvious from examination of Figures 4 to 7 that the differences betneeen the spectra of the individual olefins and between the average spectra of the various olefin rlasses are so small that
ANALYTICAL CHEMISTRY
234
VAPOR SPECTRA SPECTRAL SLIT WIDTH 17A
1
WAVELENGTH IN ANGSTROMS
Figure 8. Comparison of far ultraviolet spectra of the cyclic and noncyclic mono-olefins
VAPOR SPECTRA SPECTRAL SLIT WIDTH 1.7A
the t.otal olefin determination suggested in the previous section would include the cyclic olefins as well as the alkenes. Conjugated Diolefins. Figure 9 shows the vapor spectra of five of the conjugated diolefins. These compounds exhibit a broad, intense ( e about 20,000 to 28,000) band near 2150 A., analogous to the S - T ' h n d of t,he mono-olefins. It has a comparable vibronic structure and an integrated intensity about twice that of the concsponding band for the mono-olefins. This hand has been used for the estimation of 1,3-butadiene in Cqolefin mixtures and may have more general utility. I n addition to these broad bands, t,here are also a number of sliarp bands beIon. 2000 A. which have tieen attributed to Rydberg transitions ( 6 ) . Because these band3 arc1 sufficiently strong, sharp, and charitcteristic, they could he used to identify or estimate individual diolefins in fairly complex mixtures of hydrocarbons. The work of Jacobs and Platt ( 1 4 ) indicates, hotvever, that this sharp structure is absent in t.he solution spectra, so that it \vould be necessary to use the vapor technique for such analyses. Acetylenes and Allenes. Platt, Klevens, and Price ( 2 4 ) have measured the far ult,raviolet, spectra of n-heptane solutions of 1-octyne and 2-octyne and found a weak absorption band ( e about 150) at 2250 -1.in each cafie. They also observed a stronger band ( e about 9500) :it 1800.1. for the 2-octyne and st,eadily increasing ahsorptioii t o 1700 A . ivith a step-out at 1850 A. ( e about 2000) for I-octyne. The vapor spectrum of I-hutyne (Figure 10) is somewhat similar to P h t t ' s results for the octynes with a step-out a t 1850 .I.aiid a masinuxn at 1i20 A. ( e = G O O ) . Apparently the center of tlic itlworption txind system lies below 1700 The spectra of three alkyl-substituted ttllenes are also shown in Figure 10. The monosubstit'uted compounds, 1,2-butadiene arid 1,2-pentadiene, have :t band a t 1780 ;1. with e about 20,000 and a step-out a t 1880 A . There is also evidence of a third band below 1700 A. I n t,he spectrum of the disubstitut,e allene, 2,a-pentadiene) all of these hands are shifted 30 or 40 -1.to longer wave lengths and a new hand appears at about 2000 9. Nonconjugated Diolefins with Isolated Double Bonds. When the double bonds of these vompounds are separated by two methylene groups, thew is no interaction and the spectra resemble
SPECTRAL SLIT WIDTH L7A
A I.OO0
I
. c-c.c.c-c
cc u) 0
WAVELENGTH IN ANGSTROMS
Figure 9. Far ultraviolet spectra of vapors of C, to Cg conjugated diolefins
ZF0,WO
\ \E
K
4 8G10DO0
Cyclic Mono-olefins. The spectra of vapors of cyclopentene and cyclohexene are shown in Figure 8 along with those of cisand trans-2-butene for comparison. Carr ( 7 ) has cited the similarity of the long wave-length bands of cyclohexene and of the cis-2-alkenes as evidence for the structure of the latter, since the cyclic compound can exist only in the cis configuration. Although more data will be required to reach definite conclusions, it appears from the spectra of the parent cyclic compounds that
'7 L .,
\F
\\
E 1.4-PENTPSIENE c*c-c-c.c
k 1.5-HEXADIENE
Y
c.c-c-c-c. c
G I.Oo0 G G. I
G. I-PENTENE
c.c-c-c-c 1 m m ) 9 o o x x x ) 20 0 WAVELENGTH IN ANGSTROMS
Figure 10. Far ultraviolet spectra of vapors of I-butyne and nonconjugated diolefins
V O L U M E 27, NO. 2, F E B R U A R Y 1 9 5 5
235
those of the corresponding mono-olefins except for a twofold gain in intcnsity. Thus the spectrum of 1,5-hexadiene (Figure 10) has the same contour as that of 1-pentene except for some lose of vibronic structure and the expected increase in intensity. On the other hand, a single methylene group does not provide com-
SOLUTW SPECTRA SOLVENT: “EPTANE SPECTRAL SLIT W Q N A
AKX)
plete “insulation” of the double botids (53). Thus the absorption halid system of 1,4-pent,adiene is much broader than that of 1-pentene and the peak intensity is only about 409/, great,er. Benzenes. The spect,ra of benzene, Tetralin, and 14 alkylsubstituted benzenes are presented in Figure 11. All of thesc coinpounds have two absorption bands between 1700 and 2200 The strong allowed r-elcctron transition [‘B in Platt’s notation ( 2 0 ) ]gives a hand with maximum absorption a t 1790 -4. in the benzene vapor spectrum and a t 1840 A. in the solution spectrum. It shifts t o longer wave lengths wit,h increasing alkyl substitution. ii second t,ransitioti, which has been described variously as a forbidden “free” r-electron transition (‘La in I’latt’s notation) and as a combination of such a forbidden transition with a Rydberg-type transition of a r-electron, results iii a relatively flat absorption band about 150 A. wide between 2000 and 2200 A. for all of t8hebenzenes. The maximum absorptivity of each of these bands is essentially unaffected by alkyl suhst.it,ution as is shon-n in Table I\‘. The spread of only 15% in peak iritelisity of the ‘B band found in the present work is to be cotitraded ivith a spread of nearly 50% in earlier papers (41,25).
‘I’ahle I\’. Location and Intensity of Far Ultraviolet .ibsorption Bands of Benzenes in n-Heptane Solution IR R a n d LLa Band ____. Compound
Benzene Toluene n-Decylhenzene o-Xylene I-h1 ethyl-2-ethylbenzene 1,2.3,4-Tetrahydronaphthnlenc vi-Xylene 1-lfethvl-3-ethrlbenzene p-xylelie 1-1Iethyl-4-ethylbenzene 1,2,3-Trimethylbenzene 1,2,4-Triniethylhenzen~ 1,3,5-Trimethylbenzene I .3- Di tnetliyl-5-etIi4-lhenzene 1.2.8.5-Tptraniethylbenzene 1,2,I,j-’r,tranietli).Ibenzene AV.
Std dev. 8 5 % confidence iiiiiits of a!’.
I Figure 11.
1
I
I
Far ultraviolet spectra of henzenes
V
have length,
A.
1840
1890 I900 1920 1825 1865 1940 1945 1835 1633 1950 1980
I085 1985 1666 lCJi5
e
(center of band) G0,OOO 56,000 58,000 33.000 35,000 47,000 50,000
31,000 54,000 55,000 35,000 ;,1,000 55,000 50,000 55,000 50,000 3,000 x.000 kI,flGO
Raie length, A. 1950-2100 1990-2130 2000-2130 2030-2170 2030-2170 2050-2190 2040-2180 2040-2180 2030-2220 2030-2240 2040-2200 2040-2200 2070-2220 2100-2200 2100-2240 2090-2220
.
I’rdh, c
7,900 8,100 7,600 9,000 9,000
9,100 8,000 8,400 7,500 8,700
10,000
8,200 8,800 9,100 0,100 9 , GOO 8,iiSO 710 1390
‘i’liese batids should be useful for d(3tcrniiiiation of thc, bciizriies. Thus total berizeiies might \vel1 tic tic~torniinedat an appropriate position it1 the ‘La band regioti--for cuainple, a t 2110 A. where all the alkyl benzerie~have w r y ~ i c ~ i t ,identical ly molar absorptivities. On the other hand, t,he position oi the ‘ B band depends very largely on the number of alkyl substituents. A rough estimate of the relative concentrations of niono-, di-, and tri-, or more highly substituted benzenes could thus he mude on the basis of measurcnirnts a t wave lengths corrwponding to the average positions of the ‘B band for these classes of compounds. This scheme will not be of high accuracy, howevcr, as there is a considerable amount of overlapping of the ’ B hands for the various classes. The total alkyl benzene determination by this method will therefore be better than the tireakdowri according to number of substituents. Use of the ‘ B hands rather than the ‘Labands will tend to minimize interferences i n many rases, as these bands are among the most intense which havc: been observed for any chromophore in this region of the spectrum. Catacondensed Aromatics. The far ultraviolet spectra of a number of catncondensed polynuc.lear aromatic hydrocarbons are given in Figure 12. From an analytical viewpoint one of the most interesting features of these spectra is the fact that none of these compounds has absorption baiids in the 1800- to 2000-.4. region which are as intense as the 1B bands of the benzenes. This is in marked contrast t o the situation in the near ultraviolet, where all of the polynurlear aromatics absorb much more strongly than the benzenes a t all wave lengths. The naphthalenes have no clearly discernible bands in this region. Klevens and l’latt
ANALYTICAL CHEMISTRY
236 ( 1 7 ) have reported finding a band (lBa) a t 1G70 A., which was beyond the lower limit of the measurements used here, and another extremely weak band a t about 1894 A. The present authors obtained a barely perceptible hump in the naphthalene spectrum a t about 1900 A,, but comparable bands cannot be found in the spectra of the monomethyl naphthalenes. This hand is accentuated, however, in the spectrum of 1,2’-dinaphthyl and shifted to about 1950 A. I n this compound steric hindrance between the hydrogens of the two rings prevents their being coplanar, so that effective conjugation of the rings probably does not occur. This is reflected in the negligible shift in position of theI2200-A. (’Bb) band and in the generally strong resemblance of the 1,2’-dinaphthyI spertrum to those of theothernaphthalenes.
in chryaene, a t 2020 4. in l,a-benzanthracene, and a t 2050 A. in 20-methylcholanthrene. Saturated Hydrocarbons. The saturated hydrocarbons are, in general, transparent in this region and thus suitable for use as solvents. Purely for comparison with the foregoing data on unsaturated compounds the spectra of 2,2,4-trimethylpentane and n-decylcyclohexane are shown in Figure 13. More complete absorption spectra have been given by Pickett et al. (19) for cyclohemne and cyclopentane. Data on the short wave-length cutoffs of a number of paraffinic hydrocarbons have been given by Plntt ( 2 2 ) and hy Potts (47). GENERAL OBSERVATIONS
In general the far ultraviolet spectra of solutions of all the unsaturated hydrocarbons may be characterized as consisting of very broad, nearly structureless bands. The spectra of all members of a given class of compounds are similar with respect to both position and intensity of their absorption bands. The far ultraviolet spectra may thus be expected to be of greater value in estimating the total concentration of all members of a group of compounds with similar electronic structure than they will be for distinguishing the various members of the group. This is perhaps fortunate, since other spectroscopic techniques are often well suited to the latter job but ill adapted to the former. The far ultraviolet method may be expected to be complementary to such methods as mass spectrometry and absorption spectrophotometry in the near ultraviolet and infrared.
1,000
s
I
SOLUTION SPECTRA SOLVENT: n- HEPTANE
2@MTrnCmXANMRPIE. L
L 1o.OOo
16oom leal
I
T
1900 2ooo 2100 2m2m
WAVELENGTH IN ANGSTROMS
Figure 12. Far ultraviolet spectra of n-heptane solutions of catacondensed polynuclear aromatics
Platt (20) has pointed out that the spectra of all catacondensed aromatic compounds containing the same number of aromatic rings are similar with respect to the number and location of their absorption bands, although differences in the relative intensities of these bands will result from differences in molecular symmetry. These effects are demonstrated by the spectra of phenanthrene and anthracene. The ‘B, band corresponding to an allowed transition involving a transversely polarized excited molecule occurs a t 1855 A. for anthracene and a t 1875 A. for phenanthrene. The peak molar absorptivities are of the same order of magnitude. On the other hand, the ‘Cb band (excited molecule with quadripole moment) of anthracene is “forbidden” by symmetry considerations, while that of phenanthrene is “allowed.” Although the ‘Cb band of both compounds is centered near 2100 A., the peak intensity is only 9700 for anthracene compared to 37,000 for phenanthrene. Similarly, both chrysene and 1,2-benzanthracene have absorption bands (IC@)near 1830 A., although that of chrysene is more intense bj- a factor of over two. This band shifts 50 A. toward the red in the substituted benzanthracene, 20-methylcholanthrene. The ‘I?, band of all these compounds has a maximum intensity range of only 18,000 to 27,000 but exhibits rather large changes in position. It occurs a t 1950 A.
I
1700 WAVELENGTH IN ANGSTROMS Figure 13. Far ultraviolet spectra of two saturated hydrocarbons On the other hand, the vapor spectra of many compounds exhibit numerous sharp Rydberg lines which might be useful for the identification of individual members of a series, as was mentioned in the discussion of the spectra of the diolefins. For the most part these bands lie a t shorter wave lengths (1000 to 1700 A . ) than were covered by the present investigation. Quantitative measurements below 1700 A. may be made by k n o w photoelectric techniques-e.g , (39)-but with considerablyless convenience and probably with lower accuracy than is possible a t longer wave lengths because of the greater effect of stray long w-ave-length radiation on the phosphor-sensitized phototubes which are generally used in this region. Also no stable source of continuous radiation is known for the 1000- to 1700-A. interval, so that the many-lined spectrum of hydrogen must be used. This places
V O L U M E 27, NO. 2, F E B R U A R Y 1 9 5 5 a serious limitation on the resolution and accuracy which inay be obtained with a simple single-beam recording spectrometer. Reduction of the resulting records is also extremely tedious. Despite these experimental difficulties, the analytical potentialities of the shorter wave lengths appear promising enough to warrant exploratory work. The primary purpose of the present paper is not to present finished analytical methods for the analysis of hydrocarbon mixtures but rather to stimulate a n interest in analytical applications of far ultraviolet spectroscopy. T h e potentialities are large. However, many more spectra of pure compounds must he obtained before the utility of the method can he fully assessed, :md instruments approaching the present recording spertrophotoineters for the infrared and near ultraviolet in convenience and accurarv will be requirrd before the method can he e x p e c t d t o achieve widespread acceptance. ACKNOW LEDGXIENT
The authors acknowledge the assistance of .4lIiOld Friedman and E. L. Creamer, who obtained many of the data presented in this paper. LITERATURE CITED (1) .Iinerican Petroleum Institute, Research Project 44, “Ultra-
violet dbsorption Spectral Data,” Carnegie Institute of Technology, Serial KO.497-502. (2) Ibid., Serial Nos. 46, 48, and 50. (3) American Society for Testing Materials, Philadelphia, “=ISTJI Manual of Engine Test Methods for Rating Fuels,” p. 116, 1952. (4) Bolton, J. H., and Williams, S. E., Brit. J . A p p l . Phfjs., 4 , 6-11 (1953). (5) Carr, E. P., C h m i . Recs., 41, 293-9 (1947). (6) Carr. E. P., Pickett, L. and Stucklen, H., Recs. M o d . Phys., 14, 260-4 (1942). (7) Carr. E. P., and Stucklen, H., J . S i n . Cheni. Soc., 59, 2138-41 (1937). (8) Carr, E. P., and Stucklen. H., .J. Chem. Phys., 4 , 760-8 (1936). (9) Ibid., 7 , 631 (1939). ( I O ) Dunkelinan, I,.. and Lock, C . , J . Opt. Soc. Amer., 41, 302-4 (1961). (11) Fujioka, Y., and Ito, R.. S c i . Light, 1, 1-6 (1951). J . Chem. Phys., 22, 599-602, (12) Gary, J. T., and Pickett. L. W., 1266-7 (1954).
m.,
237 (13) Hinteregger, H. E., and Watanabe, R., J . O p t . SOC.Amcr., 43, 604-8 (1953). (14) Jacobs, L. E., and Platt, J. R., J . Cliem. Phys., 16, 1137-45 (1948). (15) Johnson, F. S., Watanabe, K., and Tousey, R., J . O p t . Soc. Amer., 41, 702-8 (1951). (16) Klevens, H. B., J. Polymer Sci., 10,97-108 (1953). (17) Klevens, H. B., and Platt, J. R., J . Chem. Phys., 17, 470-81 (1949). (IS) AIann, D. E., Platt, J. R., and Klevens, H. B., Ibid., 17, 481-4 (1947). (19) Pickett, L. W., RIuntz, 11..and LlcPherson, E. hI., J . A m , Citern. SOC.,73, 4862-5 (1951). (20) Platt, J. R., J . Chela. P h y s . , 17, 484-96 (1849). (21) Platt, J. R., and Klevens, H. B., Chem. Reos., 41, 301-10 (1947). (22) Platt, J. R., and Klevens, H. B., J . A m . Chem. Soc., 69, 3055-62 (1947). (23) Platt, J. R., and Klevens, H. B., J . Chem. Phys., 16, 832-4 (1 948). (24) Platt, J. R., Klevens, H. B., and Price, W,C., Ibid., 17, 468-9 (1949). (35) Platt, J. R., Rosoff, I., and Klevens, 13. B., Ibid., 11, 536-44 (1943). (26) Pondrum, W. L., and Robertson, W. W.,Rev. Sci. I u s / r . , 19, 561-4 (1948). (27) Potts, J. R., Jr., J . CAem Phys., 20, 809-10 (1952) (28) Powell, W. I f . , Jr., PRys. Reo., 45, 154-7 (1934). (29) Price, W.C., Ibid., 4 5 , 843-4 (1934) (30) Ibid.. 47. 444-52 (1936). (31) Price, W. C.. and Tutte, W. T., Proc. Rojj. SOC. (Loridon), A174, 207-20 (1940). (32) Romand, J., and Vodar, B., C o m p t . rend., 233, 930-2 (1951). (33) Rusoff, I. I., Platt, J. R., and Burr, G. O., J . Am. Chcm. Soc., 67, 673-8 (1945). (34) Schwetzoff, V., Robin, S., and Vodar, B., J . p h y s . radium, 13, 369-70 (1952). (35) Semenom, D., Harrison, -4.J., and Carr, E. P., J . Ciienr. Phys., 22, 638-42 (1954). (36) Smith, D. C., and JIiller, E. C., J . Opt. SOC.Amer,, 34, 130- 4 (1944). (37) Tousey, R., Johnson, F. S., Richardson, J., and Toran, S . , Ibid., 41, 696-8 (1951). (38) Wainfan, N., Walker, W. C., and Weissler, G. L., J . A p p 2 . Phys., 24, 1318-21 (1953). (39) Watanabe, K., and Inn, E. C. Y., J . Opt. SOC.Anicr., 43, 32-5 (1953). (40) Zelikoff, XI., and Watanahe, R . , I b i d . , 43, 756-9 (1053).
RECEIVED for review July 17, 1964. Accepted October 30, 1954.
Determination of Hydrocarbons in Hydrogen by a Palladium Tu be-Mass Spectrometer Method R. A. BROWN, H. B. OGBURN, F. W. MELPOLDER, and W. S. YOUNG Atlantic Refining Co., Philadelphia, Pa.
A convenient method is described for concentrating hydrocarbons in hydrogen-rich samples by pumping hydrogen through a palladium tube. By combining this procedure with mass spectrometric analyses 0 to 3?4 hydrocarbons can be determined individually with an error of 0.1% and a repeatability of 0.05%. Twenty minutes are required to obtain a hydrocarbon concentrate.
D
ETER.1.IIKSTION of hydrocarbons in hydrogen-rich mix-
tures is currently of prime importance to the petroleum industry in naphtha catalytic reforming studies. I n particular, the conclusions drawn from experiments performed in both laboratory and pilot scale equipment depend to a large extent upon accurate hydrocarbon determinations for such product, streams. For units operated without gas recycle, an error of 0.1% in butanes
in the gas analysis may mean an error of 5 t o 15% or niorc, dcpending on the severity of operation, of the total butane pioduction. This, in turn, yields an error for this component of about 0.7% in over-all material balance. The significance of this error in butane analysis 011 the gas stream is reduced in its effects hen the experiment has been conducted in equipment with gas recycle. For this case, it would cause about 1 to 1.5% error in total butane production. -4nother serious problem is that small amounts of C, to C7 comporients are not usually detected by a direct mass spectrometric analysis. T h e best way t o improve accuracy and detectability in such samples is to concentrate the hydrocarbons prior t o analysis. This concentration consists of separating hydrocarbons from hydrogen by using palladium which is permeable t o hydrogen but not t o other gases ( 4 ) . Permeability increases rapidly with temperature. Fleiger ( 3 ) used this fact to determine hydrogen in mixtures containing
