Est imat ion of Pyrrole Nitrogen in Petroleum Distillates K. B. THOMPSON, TED SYMON,
AND
CHARLES WANKAT
Universal Oil Products Co., Riverside, I l l .
In view of the influence of trace constituents on the stability of certain petroleum fractions, an attempt was made to develop a quantitative method for determining pyrrole nitrogen. Extraction of the pyrroles with phosphoric acid-acetic acid in the presence of Ehrlich reagent leads to formation of a highly colored complex which lends itself to colorimetric determination. As the pyrroles present could not he isolated for preparation of a calibration curve, 2-methylpyrrole, which has a spectrum very similar to those of pyrroles extracted from fuel oils,
R
I3;CENTLY it has been recognized that the presencc of nitrogen compounds that contain the pyrrole nucleus in distillate hydrocarbon fractions may have considerable effect 011 the properties of these materials. Thus Mapstone ( 5 ) showed that the presence of pyrroles in shale oil naphthas caused extensive gum formation. The ability of certain pyrroles to promote sediment formation and discoloration of distillate fuel oils was also pointed out by others ( 7 ) . Because the pyrroles are extremely feeble bases or even weak acils (3),they will not be found in acid titration for basic nitrogen. Thri e are some exceptions to this generality-namely, that tetraalkyl pyrroles and 2,4-dimethylpyrrole have been found to be basic (6, 8). The application of a qualitative method for detection of pyrroles employing Ehrlich reagent was not satisfactor) for quantitative work, as the mercuric chloride reagent does not remove all of the pyrroles in the extraction ( 7 ) . However, H procedure using phosphoric acid-acetic acid for extraction as developed by Chernoff ( 1 ) for indole was found to remove the pyrroles quantitatively and to give an excellent and sensitive color reaction with the Ehrlich reagent. It is felt that phenolic compounds of the type which give colored Ehrlich reaction products do not occur in distillate oils in any significant amount: but if their presence is suspected, they can be removed by washing with dilute caustic, Examination of the absorption spectra of the reaction product of Ehrlichreagentwith three different pyrroles is shownin Figure 1. There is considerable difference in the maxima, as is also shown by the visible difference in color of the solutions. Since it was not possible to isolate the pyrroles from gasoline or fuel oil to usc aq H standard, the spectra of the pyrrole-Ehrlich reagent product
was used as a reference standard. lit 570 m p Beer's law is followed closely and good reproducibility can be obtained. This method has been applied to a variety of distillate fuel oils and has given good reproducibility in most cases. It works satisfactorily for straight-run gasoline but tends to g i v e erratic results with cracked gasoline. Because the Ehrlich reaction product of 2-methylpyrrole was used as an arbitrary color standard# nothing definite can be said about accuracy when the procedure is used to determine naturally occurring pyrrolic compounds. from several fuel oils was examined; they were found to bc very similar to the spectrum of 2-methylpyrrole (Figure 2). For this reason 2-methylpyrrole was selected as a standard for the preparation of a calibration curve. APPAR4TUS A h D REAGENTS
Coleman Universal spectrophotometer, Model 11 or its equivalent. Cuvettes, 19-mm. round type. Separatory funnels, 125 ml. Volumetric flasks, 25 ml., 50 ml., and 100 ml. Pipets, 1 ml., 2 ml., 5 ml., and 10 ml. To make the phosphoric acid indicator solution, dissolve 0.4 giitni of p-dimethylaminobenzaldehyde in 100 ml. of 85% phosphoric acid. This reagent should be prepared from the purified rompound ( 4 ) and should remain colorless. Glacial acetic acid, .P. Phosphoric acid, 85% C.P. Heptane solvent. Skellysolve c' (or the n-heptane used for knock testing) is suitable. The -olvent should be essentially paraffinic and completely frep of pyrrolic nitrogen compounds. 2-Methylpyrrole, freshly distilled, 1s used in the preparation of the calibration curves as described below. This compound is :tvailable from the Du Pont Electrochemical Division, Wilmington, Del., or it can be synthesized by the method of Fischer, Hrller. and Stern ( 2 ) . PREPARATION OF CALIBRATION CURVE
The pyrrole nitrogen content of the sample is read directly irom a calibration curve relating optical density to amount of pyriole nitrogen based on the color developed by 2-methylpyrrole. The curve is obtained by applying the analytical procedure to a series of hydrocarbon fractions containing a known amount of pyriole nitrogen. To prepare a standard solution of pyrrole nitrogen, weigh out 0.5785 gram of freshly distilled 2-methylpyriole, dissolve in Skellysolve C, and dilute to 100 ml. in a 100nil. volumetric flask. Each milliliter of this solution contains I mg. of pyrrole nitrogen. -4dd suitable portions of this solution to the hydrocarbon and follow the directions given in the proredure. Plot the data on rectangular graph paper to construct R curve of optical density 1 s. pvrrole nitrogen concentration. PR0CEI)URE
I 450
I
470
490
I I
510
I
so
I
I
1
550
570
5so
I (HO
(mp] Figure 1. Absorption Spectra of Pyrrole-Ehrlich Reagent Product WAVE LENGTH
Pipet 10 ml. of the hydrocwbori sample into a 120-ml. separatory funnel and dilute to 50 ml. with Skellysolve C. Add 5 ml. of phosphoric acid indicator solution, shake vigorously for 3 minute., then add 25 ml. of glacial acetic. acid, and shake again for 1 minute. Allow the acid layer to separate and withdraw it into a 50ml. volumetric flask. Shake the hydrocarbon with 2.5 ml. of 85% phosphoric acid, add the washings to the 50-ml. volumetric flask, and make up to volume with lacial acetic acid. I n the presence of pyrroles, a deep purple coyor will result immediately At the same time that the color of the pyrrole-p-dimethyl: aminobenzaldehyde condensation is being developed, prepare a reference standard for the colorimetric measurement. (The reference standard is necessary in order to correct for the background color olving to colored materials which result from oontact of 85% phosphoric acid with the oil sample.) The reference standard is obtained by extracting the oil sample as described in the previous paragraph, except that 5 ml. of 85% phosphoric acid
1465
ANALYTICAL CHEMISTRY to within &3%: by use of the calibration curve. However, no figure can be set a8 to the accuracy of the results on unknown hydrocarbon fractionp. DISCUSSlOh
is uaed instead of the phosphoric acid indicator solution. The reference standard is diluted in exactly the same way aq the colored extract. Set the spectro hotometer a t 570 mp, turn the drum to zero optical density, pkke a PC-4 filter in the light path, and adjust the galvanometer to read zero, using the reference standard in a 19-mm. cuvette. Transfer the colored solution to a 19-mm. cuvette and determine the optical density. The color develops immediately and fades only after standing several hours. The reference standard, however, darkens rapidly; therefore the determination must be made as quickly as possible. For the best r e d t s , it is desirable that the colored extract have an optical density between 0.25 and 0.55. If thib first color measurement givee an optical density greater than 0.55, dilute a suitable aliquot with acetic acid to obtain an optical density within the optimum range. I t is necessary that the reference standard be diluted a t the same time and in identical manner. For example; samples containing from 0.002 to 0.00470 pyrrole nitrogen require 1.0 ml. of the unknown estract dilutrd to 2a ml. with aretir wid.
Table 1. Pyrrole Xitrogen. ;\Ig./Liter 0.08
0.12 0.16 0.192 0.20 0.24 0.32 0.40
The wave length of 570 nip for determination of optical density was chosen because a t this setting Beer's law is followed very clwely, whereas a t 540 mp, a maximum, there is considerable deviation. This fact, demonstrated in Figure 3, is further supported by data of the type shown in Table I, employing known samples of 2-methylpyrrole in solution in %heptane and in a pyrrole-free distillate fuel oil. I n the latter case the results are even more erratic a t 540 m p . No explanation for this behavior can be offered a t present. I n actual laboratory operation this procedure has been carried out on a large number of different fuel oils. Occasionally the usual dark reddish purple color h not developed, but instead a different shade appears. Results with such samples are probably in error, since the a s sumption that the absorption approximates that of 2-mcthylpyrrole is wrong; fortunately, such cases are rare. In Table I1 are listed some of the results of determining pylroles in a variety of distillate fuel oils. The checks are satisfactory. In compensation for the lack of knowledge of absolute accuracy, it should be mentioned that in all cases ruamined the percentage of pyrrole nitrogen iq somewhat less thari the quantity of total nitrogen minus basic nitrogen
Optical Density
n-Heptane Solvent ..____ 540 m r 570 mp
0 . 2 2 , o .23 0.31.0.31 0 . 4 1 , O . 39
...
0.14,0.13
1,'uel Oil Solverit _____. 540 inp 570 m p
...
0.19 0 ?9,0.26
0.46 0.40
,..
...
...
0.44
0.66
0.43
0.26 0.57
...
...
..
...
...
0.18 0.26 0.31 0.32 0.38 0.62 0 65
....
C 4 LCULATIONS
Pyrrole nitrogen, u t . % S where u b
=
( a ) ( c ) ( d ) x 10-4 (b)(mI. of sample)(sp.gi)
total ml. of original colored hosphoric acid extract ml. aliquot of a taken for digtion c total ml. of diluted colored solution d = mg. of nitrogen per liter of solution (obtained from calibration curve) Where the color measurement is made directly on the original 50 ml. of colored extract-i.e., m-here there is no further dilutionthe dilution factor of c / b is unity. = = =
MC ACCURACY AND REPRODUCIBILITY
Results from both known samples containing 2-methylpyrrole and unknown samples of hydrocarbon fractions are reproducible
Figure 3 .
OF
2 - MLTHYLPYRRW N I T R M N OF SOLUTION.
Sample Curve of Optical Densit) pyrrole Concentration
PER LITER us.
%Methyl-
V O L U M E 2 4 , NO. 9, S E P T E M B E R 1 9 5 2 Table TI. Fuel Oil No.
4 6 7 8
9 10
11
1467
Reproducibility of Pyrrole Determinations
Fuel Oil Type Catalytic Catalytic Catalytic Virgin Catalytic Virgin Catalytic Thermal
Crude Source
(TCC) (FCC) (FCC)
W. Texas Ill. and Mid-Cont. Wyo. and Ill.
(FCC)
U‘. Texas
(TCC)
ni.
Ill. Ill. and Ky. Mich.
Pyrrole N, Weight % A B 0,0073 0.0073 0.0075 0,0077 0.0064 0.0067 0.0030 0.0025 0.0109 0.0098 0.0020
0.0268
0,0023 0.0244
0.0034
0.0029
reaction product. Carbazole and tetramethylpyrrole are csaniples of such materials. It is possible that not all alkylated pyrroles that have a hydrogen on a carbon atom of the pyrrole nucleus will react-e.g., Fischer and Orth ( 3 ) report that 2,3,5trimethylpyrrole will not give the Ehrlich color. LITERATURE CITED
(1) Chernoff, L. H., ANAL.CHEM.,12, 273 (1940). (2) Fischer, H., Beller, H., and Stern, A., Ber., 61, 1078 (1938). (3) Fischer, Hans, and Orth, H., “Die Chemie des Pyrrols,” gp. 8,27,
Leipeig, Akademische Verlagsgesellschaft, 1934. This procedure has been used mainly for the determination of pyrroles in distillate fuels and has given satisfactory results on both virgin and cracked stocks. It has been used successfully on straight-run gasolines, but has given erratic results on cracked gasolines-Le., in some cases it has worked well, while in others the reproducibility has been poor. The procedure has not been applied to materials boiling above a temperature of 675” F. The variation in colors produced from various substituted pyrroles has been mentioned. Production of a color upon reaction with pdimethylaminobenzaldehyde occurs only with pyrroles having at least one hydrogen atom on a carbon atom in the pyrrole nuclwu-i.e., tetrasubstituted pyrroles will not give a colored
(4) Gilman, H., and Blatt, A. H., “Organic Syntheses,” Vol. 1, p. 214, New York, John Wiley & Sons, 1941. (5) Mapstone, G. E., Petroleum Refiner, 28,111 (October 1949). (6) Richter, F. P., Caesar, P. D., hfisel, S.L., and Offenhauer, R. D., paper presented at 121st Meeting. AM. CHEM.Soc., Milwaukee, Wis., iipril 1952. (7) Thompson, R. B , Chenicek, J. A., Druge, L. W., and Synion, Ted, I n d . Eng. Chem., 43, 935 (1951). (8) Van Meter, R., Bailey, C. W., Moore, R. T., Allbright, C. S., Jacobson, I. A., Jr., Hylton, V. M., and Ball, J. s., paper presented at 121st Meeting, AM CHEM.SOC., Milwaukee, Wis., .\pril
1952 RECEIVED for review February 6, 1952. Accepted June 19, 1952. Presented before the Division of Petroleum Chemistry a t the 121st Meeting of the AMERICAN CHEMICAL SOCIETY, Milwaukee, Wis.
Fluorometric Determination of Traces of Beryllium H. A. LAITINEN
AND
PEKKA KIVALO, University of Illinois, Urbana, Ill.
The present work was undertaken in an effort to simplify the procedure of Klemperer and Martin, which involves two quantitative coprecipitation steps requiring centrifugation of the phosphate precipitates. A study of the variables affecting the fluorescence readings of beryllium with purified morin was also deemed desirable. Phosphate is removed by precipitation as bismuth phosphate from nitric acid solution. Calcium, if present in large amounts, is partly removed as the sulfate. The heryllium may be gathered by precipitation with hydrous ferric oxide at pH 6 to 6.3, follow-ed by electrolytic removal of the iron at a mercury cathode. t Iternath ely, following the removal of phosphate, the solution may be electrolyzed and subjected to the acetylacetone extraction method of Toribara and Chen. By this combined method, beryllium may be determined even in samples high in phosphate without a gathering step.
K -
i \ I P E R E R and hIartin (?) have recently described a procedure for the determination of trace quantities of beryllium in biological materials. The bulk of the calcium is removed hy precipitation of the sulfate, after which the beryllium is gathered by coprecipitation with ferric phosphate. The iron is removed by electrolysis with a mercury cathode, and the beryllium is coprecipitated with aluminum phosphate. The precipitate is dissolved in an alkaline stannite solution, morin is added, and a reading of the fluorescence is made. The present work was undertaken in an effort to simplify the procedure, which involves two quantitative coprecipitation steps requiring centrifugation of the phosphate precipitates. I n view of the uncertain state of purity of commercially available morin, a study of the variables affecting the fluorescence readings of beryllium with purified morin was also deemed desirable. Inasmuch as bismuth phosphate is the only phosphate of the rommon elements which is only slightly soluble in 0.5 11’ nitric : + c i c l , itnd bcruuse it S h m s WIT- litflr coprccipitation of iron,
aluminum, nickel, or zinc, the removal of phoRphate by precipitation with bismuth has proved useful in qualitative analysis I n the present investigation, it \\a found that phosphate could be removed by precipitation with bismuth in 0.5 N nitric acid without appreciable coprecipitation of tracae quantities of beryllium. Hydrous ferric oxide formed a t a final p H of 6 to 6.8 proved to be effective as a collecting agent for beryllium. The precipitate could be readily handled by filtration from hot solution rather than by centrifugation, and after washing it could be dissolved directly from the filter. Iron was removed electrolytically on a mercury cathode. The interference of calcium which is coprecipitated with the hydrous ferric oxide is prevented by the addition of Versene (tetrasodium salt of ethylenediaminetetraacetic acid). During the course of this work, it was learned that Toribara and Chen (15)had worked out a new procedure for the separation of beryllium from biological material, using an extraction with acetylacetone. By introducing this extraction method as an alternative to the hydrous ferric oxide gathering step, it is possible to determine beryllium even in samples containing large quantities of phosphate TI ithout any gathering steps, and without centrifugation. PURIFICATION OF MORIN
The isolation and purification of morin have recently been studied by various investigators ( 2 , 6, 10). I n this work the Perkin and Pate (11)method waa chosen because of its simplicity: Twenty grams of Eastman Kodak Co. technical grade niorin was extracted for 5 hours with absolute ethyl alcohol in a Soxhlet apparatus. The residue weighed 10 grams. T o the alcohol fraction was added an equal volume of water, and the precipitated morin was filtered off. The product was dissolved in ethyl alcohol and again precipitated by the addition of water. The morin was dissolved in 30 ml. of 90% acetic acid and heated to boiling. Twenty milliliters of concentrated hydrobromic acid wm added, and the morin hydrobromide was filtered off, washed with a small amount of 90% acetic acid, transferred to a beaker, and boiled wit11 water to decompose the hydrobromide. The product was
