Polarographic Determination of Naphthalenes in Petroleum Fractions R. A. BURDETT AND R . E. GORDON Wood River Research Laboratories, Shell Oil Co., he., Wood Riser, 111 A polarographic procedure for the determination of naphthalenes in petroleum fraotions is based on the reduction of oondensed ring bicyclic hydrocarbons at high negative potentials ( - 2.51 volts 1)s. S.C.E.). Monocyclic aromatics, paraffins, naphthenes, and mono-olefins do not interfere. Higher polynuclear aromatics interfere, but can usually he removed by distillation. The total elapsed time for a n analysis is 20 t o 30 minutes.
A
N ANALYTICAL. method for the determination of naphthalenes in petroleum fractions boiling in the kerosene or light gas oil range is described. It is based on the polarographic reduction of the naphthalenes in a dioxane-water medium cantaining tetra-n-butylammonium iodide as supporting electrolyte. Alkyl monocyclic aromatics, paraffim, nsphthenes, and monoolefins do not interfere. Higher polynuclear aromatics and olefinic aromatics wherein the double bond is conjugated with the ring do interfere, but can usually be removed by distillation prior to the analysis. Total elapsed time for the actual analysis is 20 to 30 minutes. Studies of the c h e f i d composition of certain petroleurn fractions, and the evident importance of naphthalene &s a mw material in the synthesis of compounds like phthalic anhydride @), have indicated the need for a simple routine method for the determination of naphthalenes in petroleum fractions. In so far as these fractions are concerned, the physical properties of naphthalenes predispose their occurrence in admixture with paraffins, naphthenes, alkyl beneenes and, in catalyticdly cracked materids, olefins. Any practiml analytical method must, therefore, be applicable in the presence of large amounts of these components. A,survey of the literature (1, 2, 4,6, 7) and a study of the properties of bicyclic aromatics indicated the pomibility of method development d o n g the iollowing lines:
.
1. Determination of total aromatics by sulfuric acid extrsction and calculated ratio of mono t o dicyclic aromatics from specific dispersion measlirements. 2: Spectrophotometric measurements in the ultraviolet reson. 3. Formation of the picrate adduct. 4. Polarographic reduction.
As is well known, tho tendency for aromatics to form picrate adducts has been made the basis of methods for the quantitative analysis of naphthalene (i,2,4). The method of Bennington and Geddes (S), after slight modification, w&s applied to the analysis of several straight-run kerosenes with very poor results. Duplicsto runs varied by &s much as 300%. Laitinen and Wawsonek (6,7),in their study of the behavior of substituted olefins and polynuclear aromatics at the dropping mercury cathode, have indicated that good diffusion currents for naphthalene and acenaphthene could be obtained. On the basis of their diffusion current constants, they suggested the feasibility of a quantitative polarographic method for the determination of any of the compounds having good polarographic waves, in the absenec of others having similar half-wave potentials. . I n products such as kerosene and light gas oil, the boiling range is wide enough to include naphthalene, acenaphthene, a-methyl-, naphthalene, and 8-methylnaphthalene. Furthermore, Rossini and Mair (8) have shown that in a kerosene distilled from a Midcontinent crude petroleum, naphthalene and the a-and &methylnaphthalenes were the only polynuclear aromatics present. Therefore, the method if applied to petroleum products such as kerosene or light gas oil would necessarily involve the determi+tion of total naphthalenes because of the similarity in half-wave potentials. Higher polynuclear aromatics whose waves appear
CONSIDERATION O F METHODS
Since the specific dispersion for the F and C hydrogen lines is spproximately 100 units greater for bicyclic aromatics than far monocyclics, a method based upon specific dispersion figures and total aromatic8 as determined by sulfuric acid extraction might, a t first glance, appear feasible. However, the presence of largo amounts of olefins would introduce a serious error because of their coextraction with the aromatics. A suitable base figure for the saturate portion of a sample could not be obtained without a timeconsuming percolation over silica gel t o remove the aromatics and olefins. Because of the unfortunate juxtaposition of the ultraviolet absorption bands of naphthalene, and the alpha and beta isomers of methyl naphthalene, quantitative estimates of either the individual compounds or of total naphthalenes in a mixture of all three would not be practical. In addition, in complex mixtures such &s petroleum distillates, the possibility of other compounds absorbing in the same region as the naphthalenes is ever presenl and might lead t o serious error. 843
V O L U M E 19, NO. 1 1 at the bame potentials as the bicyclics would be excluded because of their high boiling points. Indene, an olefinic aromatic whose polarographic wave also appears a t the same potential, could be detected by the fact that the wave has no plateau but rises steadily until the decomposition potential of the electrolyte is reached. If, at any time, samples of dubious origin were to be analyzed, a prior distillation could be performed which would segregate the bicyclic aromatics from any of the above-mentioned interferences. Because of the inherent simplicity, accur acy, and precision of polarographic procedures, investigation of such a method was undertaken. The detailed procedure used in the study is presented below. APPARATUS AND REAGENTS
A Sargent Model XX recording polarograph was used throughout the study. The polarographic cell is shown in Figure 1. The ground-glass surface serves as the seat for the upper half of a 24/40 groundglass sleeve which contains the capillary and rubber stopper. Th+ keeps the capillary in a vertical position at all times. The stopcock a t the bottom of the unit simplifies both cleaning the cell and collection of the mercury for the determination of m values. The cell has a total capacity of 13 ml. and as little as 3 ml. of solution can be run. The two-way stopcock, described by Heyrovskf (5),permits a gentle stream of nitrogen to pass over the solution during a run. This effectively prevents any diffusion of oxygen back into the cell during the electrolysis. Constant temperature is maintained by circulating water from a constanttemperature bath through the water jacket of the cell. All known and unknown samples were weighed by means of a 2-ml. Lunge pipet. Ten-milliliter volumetric flasks were used to dilute the unknown to volume with electrolyte-solvent. For purification of the dioxane, a distillation assembly consisted of a 5-liter kettle and a 1.8 X 90 em. (0.75 inch X 3 foot) silvered, fractionating column packed with glass helices A bucket-type still head was used and the purified dioxane collected. The dioxane was prepared by refluxing commercial dioxane over sodium hydroxide for 15 hours and collecting the fraction which distilled over a t 96' to 102' C. This fraction was then refluxed for 15 hours over sodium, which had been washed with isopentane, and the portion boiling at 100.5-101.5" C. was used in polarographic work. The best method of maintaining a supply of pure dioxane was to leave the bulk of the material in the kettle in contact with sodium and to distill at one time only sufficient solvent for 2 or 3 days' work. Each time the still was started up, equilibrium was reached by refluxing for 20 minutes before the take-off relay was turned on. Then, the first 35 ml. of distilled dioxane were discarded before any was collected for use. Any surplus dioxane collected could best be kept for a few days under nitrogen in an amber bottle at -20' C. The tetra-n-butylammonium iodide was obtained from Eastman Kodak and was twice recrystallized from anhydrous ethyl acetate before use. The naphthalene solvent was prepared by percolating a sample of straight-run kerosene over a 2.8 X 240 em. column of Davison
28 to 2Wmesh silica gel and eluting with methanol. The first fractions coming through had refractive indices less than 1.4520 a t 20" C. and these cuts were combined and used as a solvent in the preparation of standard solutions of pure naphthalenes. The reasons for using this material as a solvent were similarity in solubility characteristics of the material &nd the actual unknowns, and the fact that it would serve as a check on the effect of paraffins and naphthenes on the polarographic wave and diffusion current constant. The naphthalenes were all obtained from the Eastman Kodak Company. The naphthalene and acenaphthene after recrystallization from ethanol gave melting points of 96" and 81' C., respectively. The a- and p-methylnaphthalenes were purified by triple recrystallization of the picrates from alcohol: The 01methylnaphthalene picrate melted at 141.5-142.0' C., the p-methylnaphthalene picrate melted a t 115-116' C. The pure hydrocarbons were regenerated from the picrates by extraction of an ether solution of the latter with aqueous ammonia, evaporation of the ether, and drying in vacuo for 2 hours. 15
o BETA-METHYLNAPHTHALENE 12
o NAPHTHALENE
v)
W
a 1 9
a
0 K
0
aI
I-
6
I
0 W
t
ts3 i q
3
0 0
I
2
3
4
CONCENTRATION- MILLIMOLES/ LI TER
Figure 2.
Calibration Curve for Total Naphthalenes
Electrolyte-solvent was prepared by making an 85% by volume solution of dioxane in water and adding sufficient tetra-n-butylammonium iodide to make a 0.10 M solution. The concentration of dioxane in water was fixed to give as'high a solubility for kerosene and light gas oil as possible, and, at the same time, to give a fairly sensitive response of the polarograph to small concentrations of naphthalene. METHOD USED IN EXPERIMENTAL WORK
Table I.
Calibration Curves for Different Naphthalenes
ConcenWave Compound tration, C Height,id MilliMicromoles/l.a amperes i d / C Naphthalene 2.30 8.06 3.51 2.69 8.75 3.25 3.66 12.30 3.36 1.80 6.06 3.37 Acenaphthene 1.30 4.14 3.19 0.87 2.86 3.31 1.84 5.73 3.19 2.15 6.95 3.23 a-Methylnaphthalene 1.47 4.88 3.32 2.40 7.85 3.27 1.87 6.05 3.23 2.70 8.82 3.27 @-Methylnaphthalene 1.37 4.51 3.29 1.51 4.88 3.23 2.32 7.42 3.20 1.10 3.56 3.24 a Millimoles/liter of electrolyte-solvent.
Average
rlverage Deviation from Mean
id/C
%
3.37
j=2 1
3.23
*1,3
3.27
*0.6
3.24
*0.8
Weigh the sample into a 10-ml. volumetric flask by means of a Lunge pipet. Usually not more than 0.32 gram of kerosene or 0.25 gram of a light gas oil will dissolve in 10 ml. of electrolyte solvent. Dilute the sample to the mark, mix, and transfer to the cell. Pass nitrogen through the solution for 13 minutes, insert the capillary, and scrub for 2 minutes more. Turn the threeway stopcock to permit nitrogen to pass over the solution, and record the polarogram from -1.80 to -2.30 volts. Operate the polarograph at full damping position and use a span potential of 2.0 volts. Determine the wave height by subtracting the residual current from the diffusion current a t -2.30 volts impressed potential. Place a scrubber containing 85% dioxane between the cell and nitrogen source t o avoid loss of solvent by evaporation. Immediately after the polarogram is recorded, withdraw the capillary, rinse it with acetone, and dry with a soft, absorbent tissue. Rinse the cell thoroughly with acetone and blow it dry with an air stream. EXPERIMENTAL
The capillary constants were t = 6.35 sec., m = 0.850 mg. sec.-l, mz41/6 = 1.218 mg.2/3sec.ll6, open circuit. The mercury column, h , was 60.6 em. and the potential of the internal anode
a45
NOVEMBER 1947 was -0.443 volt vs. the saturated calomel electrode. After standard solutions of the pure naphthalenes in the aromatic-free kerosene were prepared, a calibration curve for each compound was obtained by recording polarograms of different weights of each. The results are summarized in Table I.
*
The data show satisfactory precision in the development of standardization curves. The line in Figure 2 was drawn by using the average value of the data under average i d / C in Table 1. Since the calibration curves fall so close together (within 2.5%), and since the unknown would undoubtedly be composed of mixtures of naphthalenes, an average line among the four would represent not only the four individual curves, within experimental error, but also the composite curve, thereby becoming the working curve for use in the analysis of unknowns. Occasionally in polarographic analysis, examples have appeared wherein two substances exhibiting normal behavior singly, exalt or suppress each other’s wave height when electrolyzed in mixture. I n order to demonstrate the presence or absence of this phenomenon, known mixtures of the various naphthalenes were prepared in dearomatized kerosene and subjected to polarographic analysis. The observed diffusion currents were compared with the computed values based on the contribution of each component in the mixture to the total wave. Table I1 shows that the wave height of mixtures of the naphthalenes is the sum of the wave heights of each component present. Since it wm evident from the successful use of dearomatized kerosene as a solvent for the pure naphthalenes that paraffins and naphthenes did not interfere, it remained to determine whether the alkyl and naphthenyl benzenes present in the boiling range close to that of the naphthalenes would present any problem in the analysis. iZccordingly 100 ml. of a straight-run kerosene were percolated over silica gel and eluted with methanol. Cuts of 5 ml. each were takrn from the column until the first rapid rise in refractive index was noted, and thereafter a t 2.0-nil. intervals. Preliminary, semiquantitative experiments showed that the cuts between 80 and 857, through were relatively low in naphthalenes but very high in total aromatics. These cuts, referred to below as “pure desorbate,” n ere combined and were polarographicallv reduced alone and in mixture with naphthalene. The figures under calculated wave height were obtained by adding the wave height due to the bicyclics in the pure desorbate to that for the naphthalene. Results were expressed in weight per volume rather than moles, since no attempt was made to determine the exact concentration of naphthalenes in the pure desorbate. However, Table 111 does show that the alkyl benzenes normally present in a kerosene do not intei fere.
Table 11.
Wave IIeights of hlixtures of Naphthalenes
Mixture Composition
+
Naphthalene a-methylnaphthalene a-Methylnaphthalene 6methylnaphthalene 0-Methylnaphthalene acenavhthene Kaphthalene a-methylnaphthalene pmethylnaphthalene
+ +
++
Concentration in ElectroWave Height lyte-Solvent Observed Calculated ~ f i l l i m o l e s / l . Microamperes Microamperes 3.21 14.01 14.22 1.04 2.71 16.39 16.83 2.46 ’ 1.82 9.42 9.48 1.11 1 04 19.07 19.28 1.82 3.04
Table 111. Influence of Alkyl Benzenes on Wave Height of Naphthalenes Pure Desorbate in 10 MI. of Electrolyte Solvent Gram 0.1903 0.1143 0.1419 0.1611
Added Kaphthalene Mo None None 0.41 0.60
.
.
Wave Height Observed Calculated Microamperes Microamperes 2.63 ... 1.58 ... 2.95 3.03 3.87 3.81
Table IV.
Sample 1 2 3 4 5 6 7
‘
Analysis of Unknowns for Total,Naphthalenes
Weight in 10 hll. of ElectrolyteSolvent Gram 0.1618. 0,1082 0.1442 0.1612 0,1122 0.1996 0.1059 0,1020 0.0798 0.1289 0.1205 0. 1191 0.1346 0.1275
Wave Height Microamperes 10.40 7.28 9.38 10.69 5.90 10.33 10.43 9.74 8.82 15.68 9.12 8.95 8.04 7.58
Naphthalenes per Liter of EleotrolyteSolvent Millimoles 3.17 2.22 2.86 3.26 1.80 3.15 3.18 2.97 2.69 4.78 2.78 2.73 2.46 2.31
Naphthalenes per Kg. of Sample Millimoles 196 205 198 202 160 158 300 29 1 337 37 1 231 229 182 181
ANALYSIS OF UNKhOWNS
The method, as described above, was applied to several refinery samples, both of straight-run and cracked stocks. For convenience, the results are expressed in millimoles of total naphthalenes per kilogram of unknown rather than in weight per cent because of the similarity of calibration curves when expressing results on a molal basis. The data are presented in Table IV. In case conversion to a weight per cent figure is desired, it is only necessary to multiply the quantity “millimoles per kg.” by an average molecular weight for the naphthdenes and divide by 10,000. DISCUSSION
X study of the diffusion current constants in’ Table I shows t i a t , when a working curve which is the average of the four calibration curves is used, the largest possible error resulting from unequal distribution of the naphthalenes is only 2.5%. This makes the method very attractive for determining either total or single naphthalenes, since the same working curve may be used in either case. The most difficult part of the entire work was the development of a technique by which one could obtain satisfactory polarograms using the same capillary over a long period. Because of the high potentials involved, slight contamination of the capillary, which had no deleterious effect on diffusion current at mnre positive potentials, resulted in ~ ~ i i d fluctuating ly waves a t the potentials necessary for the reduction of naphthalenes. Attempts to use capillaries having high m values helped somewhat, but did not give satisfactory performance over considerable periods. Inspection, through a stereoscopic microscope, of a capillary operating in electrolyte solvent a t an impressed potential of -2.30 volts revealed continuous formation of gas bubbles 1 mm. or so above the orifice. These bubbles appeared to enlarge until suddenly expressed from the electrode by the flowing mercury, thereby causing an erratic drop. It was thought that this might result from the diffusion of the polarographic solution into the capillary followed by the subsequent volatilization of the dioxane or from the electrolytic production of minute quantities of hydrogen. In substantiation of the solvent contamination theory, ib was observed that new capillaries usually gave satisfactory waves, but. exposure of a new capillary to the solvent a t -2.30 volts for 0.5 hour or so invariably resulted in a capillary unfit for we. Several methods were studied as a means of overcoming this obstacle. Best results were finally obtained by rinsing the capillary with acetone immediately after a run, and, when the polarographic wave was erratic, by immersing the capillary in nitric acid for a few minutes while the mercury still flowed. Those capillaries lasted longest which were exposed to the dioxane for the shortest period. Therefore the capillary was inserted into
846
V O L U M E 19, NO. 1 1
the cell only a few minutes before the polarogram was recorded. Ultimately, evety capillary became so contaminated that the above treatment had no effect. In that case, the capillary was removed from the assembly and attached to a vacuum line. Immediately after the mercury was sucked out, the tip was immersed in aqua regia for 20 minutes, followed by immersion in best distilled water for 10 minutes, and finally in pure acetone for 15 minutes. With suction still applied, the capillary tip was exposed to clean air for one minute to remove the acetone, and then removed from the line. It wap then ready for use. I n every case the above treatment restored the capillary to a usable condition. Although there are undoubtedly other polarographic solvents which are more easily purified than dioxane, the latter was retained because of its excellent solvent properties for kerosenes, light gas oils, etc., a t high water-to-dioxane ratios. The half-wave potentials of naphthalene, a-methylnaphthalene, and @-methylnaphthaleneoccur at -2.51 volts us. the S.C.E., while that for acenaphthene is a t -2.60 volts us. S.C.E. The diffusion current is measured at -2.74 volts us. S.C.E. (or -2.30 volts impressed potential) which is sufficiently negative to allow for quantitative results even when large quantities of acenaphthene are present. Because of the ready solubility of oxygen in organic solvents like dioxane, the two-way stopcock shown in Figure 1 was used. Without a stream of nitrogen over the solution sufficient oxygen was bqund to diffuse back into the cell in 3 minutes and to result in a very high residual current. Ordinarily, a completely closed system would be used in such a cell and the capillary assembly itself relied upon tp exclude the air. However, it became evident early in the investigation that capillaries would have to be disassembled far more frequently than is usually the case in polarography. Therefore, a one-hole stopper was cut lengthwise through to its center, so that capillaries could be inserted by merely prying apart the split stopper and inserting the electrode. Then, the stopper with the aid of a trace of glycerol was inserted iinto the ground-glass sleeve which kept the capillary in a fixed position. A slight opening remained for the exit of the nitrogen. Acetone is an excellent wash for the capillary and cell. How-
ever, traces of it will result in erratic residual and diffusion currents, so the equipment must be blown entirely free of acetone between runs. The method, as described, can be used for petfoleum fractions in any boiling range, provided higher polynuclear aromatics are not present. Biphenyl (231, z = -2.70 volts us. S.C.E.) may also intcrfere if present in smaller quantities than the naphthalenes since, in this case, there would be no perceptible plateau between the two types of compounds. In the presence of equal or smaller quantities of naphthalenes, it can be identified by the slight plateau which occurs between the two waves, and a rough estimation of the quantities of naphthalenes and biphenyl can be made. In the presence of acenaphthene, however, the biphenyl wave appears as a single wave with this naphthalene. Kerosene and light gas oil are mentioned here because the naphthalenes and no other polynuclears appear in these products because of their boiling range. Other materials can also be analyzed by this method merely by changing the kerosene solvent to one more appropriate for the analysis. Of course, the presence of other electroreducible substances in the unknown must always be investigated and, if present, the extent of their interference must be determined. LITERATURE CITED
“Allen’s Commercial Organic Analysis,” 6th ed., Vol. 111, pp. 214-15. PhiladelDhia. P. Blabiston’s Sons and Go.. 1925. Bennington, D., aid Geddes, W. F., IND.ENG.CHEM.,A N A L . ED.,6, 461 (1934). Brown, H. T., and Frarer, J. C. W., J . Am. Chem. Soc., 64, 2917 (1942).
Colman, H. G., Gas J.,144, 231 (1918). Heyrovskg, J., “Polarographie,” p. 326, Vienna, J. Springer, 1941. Laitinen, H. A . , and Wawronek, S., J . Am. Chem. Soc., 64, 1765 (1942).
Ibid., 64, 2365 (1942). Rossini, F. D., and Mair, B. J.. Refiner Y a t u r a l Gasoline M f r . , 20,
494 (1941). R i . . c r : ~ r r .N~ x r c h 29. 1947.
Determination of Vanillin and Related Compounds after Treat ment with AIkali A Spectrophotometric Method H. W. LEMON, Department of Biochemistry, Ontario Research Foundation, Toronto 5 , Canada
T
HE effect of alkali’on the ultraviolet absorption spectrum of p-hydroxy aldehydes and p-hydroxy ketones has been reported ( 5 ) . In alkaline ethyl alcohol solutions the long-wave bands of these compounds are displaced into the high ultraviolet (328 to 370 mp) and their absorption intensities are considerably increased. I t was thought that measurement of the intensity of absorption of alkaline solutions at the maximum might be used for the quantitative determination of some of these compounds, as the absorption by most other compounds in this region of the spectrum is small. The following study was therefore undertaken in order t o develop a method for the determination of vanillin. MATERIALS AND APPARATUS
The solvent used was 9595 ethyl alcohol, refluxed with zinc dust and potassium hydroxide before distilling from all-glass
equipment. Water can be used, but the absorption characteristics are somewhat different. A solution of potassium hydroxide was prepared as required bv diluting a 4y0stock solution to 0.2% with 95% ethyl alcohol. When this solution was diluted xith alcohol to the concentration used in the alkaline sblutions, the absorption a t 353 mfi as compared with pure alcohol was small. The solution was checked frequently, and Tvas discarded if an increase in absorption was observed. A Becknian spectrophotometer was used for measuring absorption intensities; an ultraviolet accessory set is not necessary. ABSORPTION
Effect of Alkali Concentration on Absorption. The wave length of maximum absorption for vanillin in alkaline 95y0ethyl alcohol is 353 mp. To determine the amount of alkali necessary for maximum absorption at this wave length, alcoholic solutions were prepared which contained varying amounts of 0.297, potassium hydroxide and 0.4 mg. of vanillin in 100 ml. The results
