ANALYTICAL CHEMISTRY
550 thus obtained are used in the reciprocal matrix to calculate the remaining components. The optical density of the sample is determined a t the preselected wave lengths. The observed optical densities are corrected for pressure deviation from standard pressure; pressure deviation from ideal gas law; temperature deviation from standard base plate temperature; false energy contribution; and the mole fraction of the components previously determined from the mono peaks of the mass spectrum. These optical density corrections are calculated from the following equation:
Di
=
XiAi
+ ADi
where Di is the optical density of component i, X i is the mole fraction of i , Ai is the A value for component i a t wave length involved, and hDi is t,he correction of the optical density of component i due to failure to observe Beer’s lax. These corrected optical densities are then used in the calculation of the sample by using the reciprocal matrix. Solution of the reciprocal matrix will give the mole fraction of each component. The mole fraction thus determined plus those previously determined from the mono peaks of the mass spectrum for hydrogen, methane, ethane, and propane multiplied by 100 should total 100.0%. If the results deviate from lOO.O%, they are normalized in the folloiving manner: ( a ) Add to or subtract from each original value l / n of the difference between the sum and 1.000, where,n is the number of components in the sample. ( b ) Add to or subtract from each original value its own percentage of the difference between the sum and 1.000. (c) Take the averages of t,he mole fractions determined in (a) and ( b ) as final values. Note. The normal lack of accuracy may cause the calculated percentages of components present to less than 0.5% to be slightly negative. These are reported as zero. Negative values larger than 0.7% indicate an error either in calculation or in the experimental data. If the deviation of the mole percentages from 100.O~ois greater than 0.5 times the number of components, check the calculation and experimental data for errors. DISCUSSION
With the use of this method of calculation, it is necessary to obtain for each sample both the infrared optical densities a t preselected wave lengths and the mass spectrum. The equations to be used to calculate the concentration of any particular component are selected from the method that gives the greater accuracy. In general, one or more isomers in a homologous series will be more accurate by the infrared method of analysis; others, when determined by the mass spectrometers. A sample of butanes and butenes contaminated by a small amount of propane may be used to illustrate this method.
From a study of the mass spectra and infrared spectra of the components, it x a s noted that the propane and la-butane hnd high interfdrence with each other on the infrared spectrometer and very little interference on the mass spectrometer. I t was therefore decided to use the data from the mass spectrometer to analyze for propane and n-butane. Because there is little interference with the determination of 2-methylpropane (isobutane) with the infrared spectrometer, data from the infrared spectrometer were used to determine the 2-methylpropane content. In like manner, 2-methylpropene and 2-butene viere selected to be analyzed from infrared spectrometer data and 1-butene from mass spectrometer data. Analytical data obtained from this method are presented in Tables I and 11. These analyses are in agreement with the sample composition known from the synthesis of the mixtures which were analyzed. Table I11 compares the results of analyses of plant stream samples obtained by complementary infrared mass spectrometer analysis with those determined by the mass spectrometer. In samples 6 and 7 , the mass spectrometer results show propylene and 2-methylpropane in the 2-methylbutane samples, whereas the infared-mass spectrometer (IRAIS) analysps gave no propylene or 2-methylpropane. From a knodedye of the history and flow of the sample through the refinery, it, is unlikely that either sample contains 2-methylpropane and definitely should contain no propylene. I t would seem therefore that the combined analyses for these two samples are more accurate than the mass spectrometer. This method of calculation can be applied t o any gaseous inixture. The method is not restricted to the wave lengths and sensitivity coefficients presented in this paper but it is possible to use any combination of infrared optical density measurements and sensitivity coefficients to calculate the analysiq of the sample. ACKNOWLEDGMENT
The authors wish to thank the Cities Service Refining Corporation for permission to present this paper. LITER-ITURE CITED (1) Consolidated Engineering Corp., Pasadena, Calif., Computing
Manuals I and 11. (2) Crout, P. D., Trans. Am. Znst. Elec. Engrs., 60, 1235 (1941). (3) Frazer, Duncan, and Collar, “Elementary Matrices,” Chap. IV, Cambridge University Press, 1938. (4) Rubber Reserve Co., Rept. 16T-45 (Sept. 5, 1945). (5) Technical Advisory Committee, Petroleum Industry War Council, Rept. HC AC-7, Analytical Method 4-B7-TR44 (Sept. 2, 1944). (6) Texas Co., Rept. 1576 E (May 1, 1946) ; private communication. RECEIVED February 2, 1948. Presented before the Southwest Regional Meeting, Houston, Tex., December 12 and 13, 1947.
Ultraviolet Absorption Analysis for Naphthalenes NORMAN D. COGGESHALL AND ALVIN S. GLESSNER, JR. Gulf Research & Development Company, Pittsburgh, P a .
T
HE; present report describes analytical methods based on ultraviolet absorption spectra, developed for the determination of naphthalene, a-methylnaphthalene, and p-methylnaphthalene in hydrocarbon mixtures boiling in the kerosene range. Despite t,he rather close similarity and proximity of the absorption bands of these compounds, it has been found practical to utilize them for quantitative analysis wnrk. Such a method is feasible because of the nature of the ultraviolet absorption of the other types of compounds found in such samples. The semiempirical theory of electronic oscillations ( 2 , S), together with the rather wide range of data now available for various classes of hydrocarbons, allows a reliable prediction to be made of the ultraviolet absorption of definite classes of hydrocarbons in the different, spectral regions.
Paraffins, naphthenes, mono-olefins, and nonconjugated diolefins are known to possess no appreciable absorption a t wave lengths longer than 200 m,x (3). The onset of absorption for conjugated diolefins is near 230 In,u and is continuous for shorter wave lengths. Blkyl benzenes and polycyclic aromatics possessing only one benzene ring-Tetralin, for example-have strong absorption in the characteristic benzene ring region, 250 to 280 mp. Mononuclear aromatics with unsaturated side chainsstyrene, for example-may possess additional bands around 290 mp if the benzene ring and an olefin group are conjugated. For all these classes, the intensity of absorption is generally decreasing very rapidly with increasing wave lengths in the vicinity of 300 m;J.. Thus they may contribute relatively little to the absorption in the region of the characteristic naphthalene
V O L U M E 2 1 , NO. 5, M A Y 1 9 4 9
551
A method, based on the ultraviolet absorption of the naphthalenes, is described for the analysis of hydrocarbon mixtures boiling in the kerosene range for naphthalene, a-methylnaphthalene, and P-methylnaphthalene. For the lower boiling cuts a method of analyzing for naphthalene alone is described. This utilizes a system of correcting for the background absorption due to other unsaturated compounds present. For the higher boiling cuts containing all three naphthalenes the method is based on data talcen at three separate wave lengths. The accuracy obtainable under routine conditions is satisfactory; the average errors are less than 0.3Yo of total sample. For proper utilization the cuts must be made between definite temperature limits. Results of accuracy tests and tests made on synthetic samples are given.
rials. This treatment, is widely used for the treat'ment of samples prior to an ult'raviolet analysis for the Cs and lighter aromatics. It consists of agitating the solution of sample in iso-octane in a strong and potassium hyaqueous solut,ion of droside for some 10 to 20 minutes a t room temper,iture.
I)nnds, 300 to 330 mp, Certain po]ycyc]ic compourlds such i,s acenaphthene and fluorene possess appreciable absorptioll in this region but fortunately have boiling points outside the range of the samples considered here. The method is a straightformrd one based on Beer's law of abeo1,ption given in Equations 1 and 2: .
I = I., x lo-
