The assistance of R. C.

Cox in writing
the computer program, W. L. Senn, Jr.,
in obtaining the NMR spectra, and J. S.
Table VIII. Nitrogen Compounds Identified in Product from Catalytic Hydrogenation of Quinoline by Hydrogen
Transfer from Diluent

GC
peak
No.

2
3
4
5
6
7

8
9
10
11
12

Compound identified Wt., yo

Aniline
12.3
o-Toluidine
5.3
&Ethylaniline
1.2
N-Ethyl-o-toluidine
0.3
o-Propylaniline (and other
4.2
compounds)
Quinoline
47.8
2-Methylquinoline and
14.1
1,2,3,4tetrahydroquinoline
3-Methylquinoline
1.3
Indole
4.4
2-Ethylquinoline
4.4
2-Isopropylquinoline
2.6
Dimers and codimers of
2.1
partially hydrogenated
quinoline, alkylquinolines and indoles
100.0

Ellerbe and J. A. Hodgeson in obtaining

some of the luminescence spectra is
gratefully acknowledged. We also
thank J. B. Zachry for supplying the
sample from denitrogenation studies
and T. P. Hawes for assistance in obtaining some gas chromatograms.
LITERATURE CITED

(1) Adler, T. K., ANAL. CHEM.34, 685

(1962).
(2) Ball, J. S., Rall, H. T., Proc. A m .
Petrol. Inst. Sect. 111 42, 128 (1962).
(3) Becker, R. S., Ph.D. thesis, Florida
State University, Tallahassee, Fla.,
1955.
(4) Chen, M., Bibliography of Phosphorescent Molecules, American Instrument Co., Inc., 1960.
(5) Drushel, H. V., Sommers, A. L.,
Cox, R. C., ANAL. CHEM.35, 2166
(1963).
(6) Ermolaev,V. L., Opt. Spectry. (USSR)
English Transl.) 11, 266 (1961).
(7) Ermolaev, V. L., Kotlyar, I. P., Ibid.,
9, 183 (1960).
(8) Gurinovich, G. P., Sevchenko, A. N.,
Solovev, K. N., Ibid., 10, 396 (1961).
(9) Hartung, G. K., Jewell, D. M., Anal.
Chim. Acta 26, 514 (1962).
(10) Heckman, R. C., J . Mol. Spectry.
2, 27 (1958).
\ - - - - I

(11) Lewis, G. N., Kasha, M., J . A m .

Chem. SOC.66, 2100 (1944).
(12) Lewschin, W. L., 2. Physik 27, 368,

382 (1931).
(13) Lijinsky, W., Chestnut, A., Raha,
C. R., Chicago Med. School Quart. 21,
49 (1960).
(14) McClure, D. S., J . Chem. Phys. 17,
905 (1949).
(15) Parker, C. A., Rees, W. T., Analyst
87, 83 (1962).
(16) Schoental, R., Scott, E. J. Y., J.
Chem. Soc. (London)1949,1683.
(17) Shablya, A. V., Terenin, A. N., Opt.
Spectry. 10,324 (1961).
(18) Shimada, R., Spectrochzm. Acta 17,
14 (1961).
\----,.

(lG)Ibid., p. 30.

(20) Supplemental DBta Sheet to Aminco

Bulletin No. 2278, American Instrument Co., Inc., Silver Spring, Md.,
Atxi1 1959.
(21)Van Duuren, B. L., ANAL. CHEM.
32, 1436 (1960).
(22) Van Duuren, B. L., J . Org. Chem. 26,
2954 (1961).
(23) Yoshida, T., Igaku Kenkyu 27,
443 (1957); C . A . 52, 10721i (1958).
RECEIVED
for review February 18, 1965.
Accepted October 7, 1965. Presented at
the Symposium on Sulfur, Nitrogen, and
Oxygen Compounds, Division of
Petroleum Chemistry, 149th Meeting,
ACS, Detroit, April 1965.

Isolation and Identification of Nitrogen

Compounds in Petroleum
H. V. DRUSHEL and A.

L. SOMMERS

Esso Research laboratories, Humble Oil & Refining Co., Baton Rouge, l a .

b A separation scheme for the isolation of nitrogen compounds from

petroleum was devised which provides
fractions of specific chemical classes.
Separation was achieved by use of
solid-liquid chromatography, chemical
extraction, and gas chromatography,
Included among the extractants were
sodium aminoethoxide in ethanolamine
and 7270 perchloric acid which were
used to isolate a variety of weakly
acidic compounds. Characterization
of the fractions from gas chromatography relied upon use of the usual
infrared, ultraviolet, and mass spectral
methods as well as sensitive fluorescence and phosphorescence techniques.
Application of this scheme to a light,
catalytic cycle oil resulted in identification of pyridines, quinolines, pyrindines, cyclopentaquinolines, indoles,
carbazoles, pyrroloquinones, phenols,
and other hydroxy compounds. Many
of these compounds are believed to be
responsible for the color, odor, and
gum-forming characteristics of heating
oils and other related petroleum
fractions.

70821

in petroleum
adversely affect many of the
important refining processes. They are
believed to reduce the activity of
cracking or hydrocracking catalysts because of their polarity and basicity.
It has also been suspected that nitrogen
compounds are to a great extent involved in gum formation, color formation, odor, and the poor storage properties of fuels. To overcome the deleterious effects of the nitrogen compounds
it is helpful to know the various types of
compounds that exist in petroleum
throughout different phases of refining.
Numerous studies have been made on
the nature of nitrogen compounds in
petroleum and several separation and
identification schemes have been reported. Hartung and Jewell (6) used
alumina adsorption, perchloric acid
extraction, and spectrophotometric examination to identify indoles, carbazoles, phenazines, and dibenzofuran
in a hydrogenated, catalytically-cracked
furnace oil. The same authors also
reported (7) the use of zinc chloride and
ferric chloride as complexing agents for
ITROGEN COMPOUNDS

the isolation of nitriles from a hydrogenated furnace oil. Carbazole was

identified in Wilmington, Calif., petroleum by Helm and coworkers (9) using
distillation, adsorption, chemical treatment, gas chromatography, and spectrometry. As a means of identifying
nitrogen compounds in gas chromatographic effluents Thompson and coworkers (18) developed a catalytic
hydrogenation micromethod in which
the nitrogen free products are identified.
Snyder and Buell (16, 17) have used
linear elution adsorption chromatography and ion exchange resins to isolate
basic nitrogen compounds, carbazoles,
and other nitrogen compounds from
gasolines through cracked gas oils.
They used ultraviolet spectrophotometry to estimate indoles, carbazoles, and
benzcarbazoles in the isolated fractions.
Sauer and coworkers (15) applied
mass spectrometry to estimate carbazoles, indoles, pyrroles, pyridines, and
quinolines in concentrates isolated from
heating oils. La Lau (12) used mass
spectrometry to identify and estimate
the quantity of pyridines, quinolines,
VOL. 38, NO. 1, JANUARY 1966

19

and carbazoles in petroleum distillates

of unspecified processing history. Dinneen and coworkers (4) described a detailed study of nitrogen compounds in a
shale-oil distillate. Separation techniques involved use of adsorption on
Florisil, molecular distillation, and thermal diffusion.
One method used by Helm and coworkers (9) for the separation of carbazole from Wilmington crude involved
treatment of a narrow boiling fraction
with sodium amide in liquid ammonia.
A precipitate was formed upon reaction
of the weakly acidic carbazole with the
sodium amide. The weakly acidic
character of carbazoles was used by
Bender and coworkers (1) for the fluorescent detection and spectrophotofluorometric characterization and estimation of carbazoles and polynuclear
carbazoles separated by thin-layer chromatography. By use of a basic system,
methanolic tetraethylammonium hydroxide in N,N-dimethylformamide, to
form the anion they were able to obtain
flaorescence a t higher wavelengths for
easier characterization.
I n the present work, use is made of
sodium aminoethoxide in ethanolamine
to extract weakly acidic compounds
from a cracked petroleum fraction.
In this system the weakly acidic materials in anionic form are soluble while
the extractant itself, being very polar,
is immiscible with the petroleum fraction. Extraction of indoles and carbazoles from other weakly acidic compounds in this fraction was accomplished
by use of 7201, perchloric acid (6).
Other phases of the separation scheme
follow a more conventional procedure.
Identification was made by combining
gas chromatography with various spectral techniques, including fluorescence
and phosphorescence ( 5 ) . The sparation was followed semiquantitatively.

3.0 KILOGRAMS
0.0046 WT. % N

SILICA GEL CHROMATOGRAPHY

I
WATER EXTRACTION
I

20

ANALYTICAL CHEMISTRY

PENTANE ELUATE
0.0000 WT. % N
0.192 WT. % s'

(WATER DISPknnmn'
Lnnuau,

(NOT STUDIED FURTHER)

RAFFINATE (IN BENZENE)

AQUEOUS HC1 EXTRACTION

I
RAFFINATE, EXTRACTED WITH AQUEOUS N a ~ C 0 3

EXTRACT, IEWRALIZED
WITH NaOH AND EXTRACTED
INTO HEXANE

I
EXTRACT ACIDIFIED WITH HCl
AND EXTRACTED INTO HEXANE

RAFFINATE, ,EXTRACTED
WITH AQUEOUS NaOH
BASIC NITROGEN

FRACTION B
CARBOXYLIC ACIDS
(COLORLESS)
0.04 g.

EXTRACT, ACIDIFIED
WITH HC1 AND
EXTRACTED INTO HEXANE

(LIGHT BROWN)

RAFFINATS EXTRACTED
WITH SODIUM AMINOETHOXIDE
IN ETHANOLAMINE
FRACTION C
PHENOLS
(LIGHT YELLOW)
0.65 g.

EXTRACT, WATER ADDED TO

HYDROLYZE AND EXTRACTED
WITH BENZENE

RAFFINATE
(NOT STUDIED FURTHER)

I
RAFFINATE (AQUEOUS), ACIDIFIED
WITH HC1 AND EXTRACTED WITH HEXANE

EXTRACT (BENZENE), EXTRACTED

WITH 72% H a 0 4

RAFFINATE

EXPERIMENTAL

Solvents. The n-hexane, 2,2,4-trimethylpentane, carbon tetrachloride,

and carbon disulfide were all spectroquality solvents except the n-hexane
used to rinse the silica gel column
which was Phillips pure grade.
Phosphorescence studies were made
using a 50-50 mixture (by volume) of
methylcyclopentane and methylcyclohexane which forms a glass a t liquid
nitrogen temperatures. Each of these
solvents was percolated through a
silica gel column t o remove traces of
luminescent impurities.
Inorganic compounds used in the
extractions were c. P. grade.
The ethanolamine was Eastman white
label grade and was used as received.
Instruments. The gas chromatograph was an F & M Model 500
with a 10% SE-30 on Chromasorb W
column. Trapping was done with
small vials with serum cap (inlet and
outlet hypodermic needles were inserted through the serum cap). A

BENZENE-METHANOL ELUATE

EXTRACT

(DARK SOLUTION)

I
RAFFINATE
(DISCARDED)

EXTRACT, WATER ADDED

TO HYDROLYZE AND
EXTRACTED INTO HEXANE

FRACTION G
OTHER ACIDIC
PnUPOUNDS
(VERY
,.-RY DARK BROWN)
I
0 . 4 z.

PRECIPITATE
(NOT STCQIED FURTHER)

I
I

FRACTION D
INDOLES & CARBAZOLES
(LIGHT BROWN)
0.35 g.

Figure 1.

Separation scheme

(Light catalytic cycle oil, Baton Rouge refinery)

special heated line with hypodermic

needle socket was attached to the outlet
of the gas chromatograph.
Fluorescence and phosphorescence
spectra were obtained with an AmincoKiers spectrophosphorimeter with fluorescence attachment.

Infrared spectra were obtained with

a Perkin-Elmer Model 221 infrared
spectrophotometer. A Perkin-Elmer
microcell was used to obtain spectra of
the fractions in CClr or CS2 solution
us. a variable path-length cell containing
only the solvent in the reference beam.

+
-

PYRIDINES

QUINOLINES 4CYCLOPENTA4
DIHYDROCYCLoPYRINDINES 4
NTAQUINOLINES

30

40

10

20

TIME (MINUTES)
Figure 2.

Gas chromatogram of fraction A (basic nitrogen compounds)

Ultraviolet spectra were recorded on

a Cary Model 14M Spectrophotometer.
Separation Scheme. A diagram of
the complete separation scheme is
given in Figure 1 for the purpose of
identifying the fractions and summarizing the details of the scheme
discussed below. The weight of each
fraction, where known, is included in
the diagram. Each fraction is numbered for the purpose of simplicity in
referring t o it in the following sections.
Most workers have used alumina for
the initial concentration of nitrogen
compounds from petroleum (6, 9). We
have found that alumina tends to promote reactions a t the surface involving
the nitrogen compounds and often retains a portion which cannot be desorbed with the most polar eluents.
This behavior of alumina is probably
related to the rather high acid strength
of the active sites on the surface. Some
of this difficulty can be overcome by
adding water to the alumina to deliberately level the acidity of the surface.
This approach has been used by Snyder
and Buell (17) to prepare the alumina
used in their linear elution adsorption
chromatographic (LEAC) separation of
nonbasic nitrogen types from petroleum.
Silica gel, on the other hand, does not
contain strong acid sites and was chosen
for the initial concentration of the
nitrogen compounds. A 1 gallon sample of light catalytic cycle oil having a
boiling range of 400-620' F. was
selected for this study. As shown in
Figure 1, the oil was percolated slowly
through a column (approximately 2
inches in diameter and 5 feet in length)
of freshly-activated silica gel. Addition of the sample to the column was
followed by elution with n-hexane to
remove the bulk of the nonadsorbed
hydrocarbons from the column. Desired nonhydrocarbon material which
remained on the column was then
eluted with a 50-50 benzene-methanol
mixture. Very little color remained
on the silica gel indicating rather complete displacement of the adsorbed compounds. The eluate was extracted with
water to remove the methanol thus

leaving a benzene solution of the nonhydrocarbon concentrate.

Basic nitrogen compounds were extracted from the concentrate with
aqueous hydrochloric acid, released by
addition of aqueous sodium hydroxide,
and extracted into spectro-grade nhexane. During each of these extractions care was taken to eliminate oxidation by flushing the void space in the
separatory funnel with nitrogen. The
basic nitrogen compounds were recovered by slowly evaporating the nhexane solution under a stream of
nitrogen while heating with a hot-water
bath.
The raffinate from the aqueous
hydrochloric acid extraction was extracted with aqueous 10% sodium
carbonate to remove carboxylic acids.
The free acids were released with
aqueous hydrochloric acid, extracted
into n-hexane, and recovered by evaporation under nitrogen in a manner
similar to that used in the recovery of
basic nitrogen compounds. The very
small carboxylic acid fraction was not
characterized further.
After removal of the carboxylic acids
the raffinate was extracted with aqueous
10% sodium hydroxide to remove
phenols and other weak acids. The
extract was acidified to release the
phenols which were extracted into nhexane and recovered by. evaporation
of the solvent.
The raffinate from the sodium hydroxide extraction was extracted with
sodium aminoethoxide in ethanolamine
(metallic sodium dissolved in ethanolamine). Very weakly acidic compounds
(indoles, carbazoles, etc.) were hydrolyzed by merely adding a large
volume of water. Compounds released
by hydrolysis were extracted into benzene and the benzene solution was in
turn contacted with 72% perchloric
acid to extract indoles and carbazoles
(6). The black precipitate, reported
(6) to be phenazines, which remained
after the perchloric acid treatment was
not studied further. The indoles and
carbazoles in the perchloric acid extract were released by addition of water

and the hydrolyzed compounds were

extracted into n-hexane and recovered
by evaporation of the solvent. A very
significant fraction was obtained by
acidification of the hydrolyzed sodium
aminoethoxide extract. Compounds released upon acidification were extracted
into n-hexane and recovered, as before,
by evaporation of the solvent in a
stream of nitrogen.
The nitrogen contents of the various
isolated fractions and raffinates were
not determined. Small sample sizes
and low nitrogen levels for certain
raffinates would have made such analyses
difficult. However, the sum of all
the nitrogen in fractions A , D,and G
(assuming an average mol. wt. of 185
and, therefore, 7.6y0nitrogen) accounts
for only a little over half of the nitrogen
originally present. The remainder of
the nitrogen would be found in the
several raffinates which were not studied
further. Nitrogen compounds not extracted probably belonged to a class of
nonbasic compounds which were too
weakly acidic to be extracted with
sodium aminoethoxide. Some of the
nitrogen compounds-e.g., the indolesmay have formed polymeric structures
difficult to extract.
Procedure for Trapping and Spectral Identification. About 30 pl. of

the isolated fraction was injected into

the gas chromatograph and fractions
were collected as determined from a
preliminary chromatogram of the
sample. Three repetitive chromatographic separations were made and
the trapped materials were combined to
provide enough sample for the various
spectral techniques applied.
Infrared spectra were obtained first
as this required the highest concentration of the trapped material. The solution (CS2 or CC14)after infrared analysis
was returned to the vial and the solvent
was evaporated in a stream of nitrogen.
Approximately 0.5 ml. of a 50-50 mixture of methylcyclopentane and methylcyclohexane was added to the vial and
the solution was transferred to a quartz
tube to obtain phosphorescence a t liquid
nitrogen temperatures (in a quartz
VOL. 38, NO. 1, JANUARY 1966

21

/\FRACTION A- 12

FRACTION A - 6

FRACTION A - 2

500

400

I
3600

3400

WAVELENGTH (np)

3200

I"
I
3000

280

FREQUENCY (CM. -l)

Figure 3. Phosphorescence spectra of some trapped GC

fractions of fraction A (basic nitrogen compounds)

Figure 4. Infrared spectra of some trapped GC fractions

of fraction A (basic nitrogen compounds)

Excitation wovelcngthr shown

Dewar flask). After recording the

phosphorescence spectra additional solvent was added in order to fill a 1-cm.
quartz cuvette. Both fluorescence and
ultraviolet spectra were then obtained
using this solution. Mass spectra were
obtained by evaporating the solutions
to a concentrate and allowing a portion

Table 1. Lifetime of Phosphorescence

of the Compounds in Trapped Fractions from Isolated Fraction A, Basic
Nitrogen Compounds

Fraction No.

Mean lifetime

A-1
A-2
A-3
A-4
A-5

(set.).

1.7
0.6
0.8
0.6
0.9
0.9
0.8
0.9
0.9
1.0
1.4
1.3
1.4

A-6

A-7
A-8
A-9
A-10
A-11
A-12
A-13

Reference compounds
2-Methylpyridine
1.3
PMethylpyridine
1.7
2,4,6-Trimethylpyridine
1.8
2,2;Bi yridine
1.0
QumoEne
0.8
a A11 values detd. in PK solvent
~~~

22

~~

ANALYTICAL CHEMISTRY

to evaporate on a solids inlet dipper.

Thus, all spectral data were obtained
with only the few milligrams of material
trapped in each fraction.
DISCUSSION OF RESULTS

Basic Nitrogen Compounds (Fraction A). Compounds in Fraction A

were separated according to boiling

point by trapping from the gas chromatograph (Figure 2). Ultraviolet
spectra of all of the fractions showed a
band centered around 275 mp with fine
structure a t 262, 270, and 285 mp.
This absorption is typical of alkyl
substituted pyridines. On the basis of
retention time quinoline was expected
to appear in Fraction A-4, but typical
quinoline absorption near 320 mp was
extremely weak in this fraction as well
as Fraction A-5. Quinolines (substituted) were present in Fractions A-6
and A-13 but they were somewhat obscured by pyridine absorption.
Fluorescence spectra provided very
little positive information for identification purposes. Acridines, which normally show strong fluorescence, were
not detected. Pyridines and quinolines,
detected in the ultraviolet spectra,
would not have been observed in the
fluorescence spectra because their fluorescence is very weak. However, Fractions A-9 through A-13 showed fluorescence near 400 mM with fine structure.

Although the emitting species were not

identified they were probably more
complex than pyridine or quinoline.
Pyrrolo rings fused to these basic
nitrogen compounds could increase the
wavelength of emission to 400 mp as
in the azaindoles (5).
Phosphorescence spectra provided
more specific information than the
fluorescence spectra. Substituted pyridines or primary or secondary aromatic
amines were indicated by broad emission
bands around 430 to 450 mp in fractions
A-1 through A-3. These types persisted through fraction A-8. Some
quinolines were found in fraction A-1,
presumably because they were carried
over from previous runs during the
multiple trapping sequence (5). Fraction A-4 showed the first evidence of
quinolines inherent to the sample.
Fractions A-5 through A-9 yielded
spectra most typical of quinoline because the pyridines present have very
weak emission. Certain unidentified
2- or 3-ring structures other than simple
alkyl substituted quinolines were indicated in fractions above A-9. The
unusual nature of the fractions above
A-9 (see fraction A-12 in Figure 3)
could be explained on the basis of difunctionality. Compounds containing a
fused pyrrole nucleus such as azaindoles
or azacarbazoles could enhance the
phosphorescence and increase the emission wavelength as observed for the
latter fractions. Some typical phos-

phorescence and excitation spectra are

shown in Figure 3. Mean lifetimes for
phosphorescence agree well with those
for pyridines and quinolines (Table I).
SUMMARY OF ELECTRONIC SPECTRAL DATA

Combined electronic spectral data

indicate pyridines to be the major class
of compounds present in Fractions A-1
through A-8.
Aromatic amines may be contributing
to the broad phosphorescence a t 430 m,u
(pyridines are normally very weak).
Although quinolines are present in
fractions A-4 and higher (by phosphorescence) they are mixed with other types
(as indicated by ultraviolet absorption).
No acridines are present.
Complex structures other than pyridine, quinoline, or acridine are indicated,
particularly in the latter fractions.
Mass spectral data showed that pyridines predominated in fractions A-1
through A-9 (Table 11). The quinolines
did not become significant in amount
until fraction A-7 or A-8 and higher as
was indicated by phosphorescence.
These data indicate that the C1 and
C2-substituted quinolines are present in
much higher concentration than the
parent compound, quinoline. Although
seldom reported ( I 3 ) , two series of
compounds, possibly pyrindines and
their dihydroderivatives seem to be
evident in intermediate fractions. Two
related series of compounds, the cyclopentaquinolines and their dihydroderivatives, may be contributing to the
last two or three fractions. The ranges
in retention time over which these
classes of compounds appeared are
shown in Figure 2. It appears that
only the more stable aromatic heterocyclic nitrogen compounds predominate.
More labile compounds such as 1,2,3,4tetrahydroquinoline would have partially decomposed during catalytic
cracking (because of C--N bond cleavage). Any tetrahydroquinoline originally present (or produced from quinoline by hydrogen transfer) could have
been converted to moderately stable
aromatic amines during catalytic cracking according to the mechanism reported
previously ( 5 ) . This is consistent with
the nature of the phosphorescence of the
first three fractions which was more
characteristic of substituted anilines
than of pyridines.
The aromatic hydrogen out-of-plane
bending region of the infrared spectrum
provides some information regarding
the type of ring substitution (Figure 4).
In every fraction except A-3 there are
numerous bands indicating several compounds or isomers in each fraction.
The width of most of the GC peaks in
contrast to fraction A-3 (which is
sufficiently narrow to be a single compound) also attests to this fact. Cook
and Church ( 3 ) and Karr and coworkers

Table II.

Summary of Mass Spectral Data for the Basic Nitrogen Compounds

Per cent
of
original
Frac- sample
tion ( X 109
A-1
0.36
A-2

1.92

A-3

0.61

A-4

1.39

5
9
5
9
5
9
5
9

A-5

0.93

A-6

0.60

11
5
9
11
5

0.53

11
5

A-7

11
A-8

0.71

Data for C,Hz,- ,N series

Largest
Relative
mle
intensity
149
117
163
(121)
131
163
131
177
145
129
177
145
143
177
162)
143
191
162)
175

100

19
43
(10;)
100

11
100
8
14
100

14
5
58
(100)

29
57

(100)

10

143
191

5
96

159
157
205

24
25
51

A-9

1.49

9
11
5

u;;)

(100)

11
7

(157, 171)
217

(54, 19)

11
11
13

(171, 185)
(185, 199)
(225, 239)

(61, 39)

11
13

(199, 213)
239

(45, 28)

15
11
13

223
(213, 227)
(239, 253)

(56, 24)
(100, 63)

15

(223, 237)

(27, 14)

7
A-10

0.91

A-11

0.68

A-12

A-13

1.01

1.17

12

10

(59, 29)
(54, 39)

100

C5-Dihydrocyclopentaquinolines
or

C4-Tetrahydroacridines

10

(IO) have systematically studied the

pyridines and quinolines, respectively,
and have presented spectra-structure
correlations for the out-of-plane bending
region. On the basis of these correlations, fractions A-1 through A-9 show
evidence of two and three adjacent
hydrogen atoms on aromatic rings
(either carbocyclic or heterocyclic).
This is manifested by the most intense
bands appearing near 785 and 810 cm.-.
Thus, mono- and disubsbitution of pyridine nuclei and disubstitution (one on
each ring) of the quinolines is most probable. Only fraction A-1 out of this
group of fractions shows any evidence
of a preponderance of four adjacent
hydrogen atoms (750 cm.+) as in 2-

Identification
Cs-Pyridines
Pyrindine?
C6-Pyridines
(fragment ion)
C1-Pyrindines?
C6-Pyridines
C1-Pyrindines?
CrPyridines
Cz-Pyrindines?
Quinoline
CrPyridines
CrPyrindines?
C1-Quinolines
CT-Pyridines?
(fragment ion)
C1-Quinolines
Cs-Pyridines
(fragment ion)
C3-Tetrahydroquinolinesor
C4-Dihydropyrindine
C1-Quinoline
Cs-Pyridine
(fragment ion)
C4-Tetrahydroquinolinesor
Cs-Dihgdropyrindines
C8-Pyrindines?
CrQuinolines
Cg-Pyridines
(fragment ion)
C6-Tetrahydroquinolinesor
C6-I)ihydropyrindines
C2-and C3-Quinolines
Ce-Tetrahydroquinolinesor
CrDihydropyrindines
C3- and C4-Quinolines
C4- and C5-Quinolines
C4- and Cs-Dihydrocyclopentaquinolines or
C3-and C4-Tetrahydroacridines
Cs- and C6-Quinolines
C4-Cyclopentaquinolines?

Cg- and C,-Quinolines

C5- and C6-Dihydrocyclopentaquinolines or
Cr and Cs-Tetrahydroacridines
C4 and Cs-Cyclopentaquinolines?

alkylpyridines. On the other hand,

fractions A-10 through A-13 all have
strongest absorption near 750 em.-
suggestive of the presence of an unsubstituted fused benzene ring. Certain quinoline-rich fractions (A-7
through A-9) elrhibit evidence for
isolated aromatic hydrogen atoms indicative of alkyl groups in the 3-, 6-, 7positions or of dialkyl substitution on
either carbocyclic or heterocyclic rings.
The C-H
and X-H
stretching
regions of the infrared spectrum were
particularly enlightening. Most of the
lower-boiling fractions showed strong
N-H absorption typical of aromatic
amines. This absorption appeared as a
doublet a t 3380 cm.-l and 3420 cm.-
VOL. 38, NO. 1, JANUARY 1966

23

TIME (MINUTES)

Figure 5. Gas chromatogram of fraction D, indoles and carbazoles

WAVELENGTH

Figure 6.

A
B
C
D

24

ANALYTICAL CHEMISTRY

(mp)

Fluorescence, phosphorescence, and excitation spectra of trapped fraction D-5

Excitation spectrum (for fluorescence)
Fluorescence spectrum (excitation a t 275 mp)
Phosphorescence spectrum (excitation at 275 mp)
Phosphorescence spectrum (excitation a t 400 mp)

(one substituent on each ring) or, in the

latter fractions, trisubstituted.
The last three fractions show evidence
for unsubstituted carbocyclic rings.

Most of the fractions show a doublet

(aromatic C-H stretching) near 3030
crn.-l and 3070 cm.-' typical of substituted pyridines as pointed out by
Heinert and Martell (8). Fraction A-5
has only a singlet near 3030 cm.-' and
rather strong N-H absorption a t 3380
and 3420 cm.-l which could be attributed to substituted aromatic amines.
Exceptionally interesting is the fact
that several fractions (see fractions A-1
and A-10 in Figure 4) show N-H
stretching frequencies (3485 cm.-l)
corresponding to pyrrolic compounds.
This constitutes some direct evidence
for fused pyrrolo rings in the latter
fractions.
SUMMARY

OF

Fraction B (Carboxylic Acids) and

Fraction C (Phenols). These two

fractions were not studied in detail.

Fraction C, the phenols, contained
fewer compounds or isomers than the
basic nitrogen fraction ( A ) . A gas
chromatogram of the phenols showed
fair separation between components
(about 12 major compounds were indicated).
Indoles and Carbazoles (Fraction

D ) . The specificity of the separation

scheme is demonstrated in the two
well-resolved groups of peaks in the
gas chromatogram of Fraction D
(Figure 5).
Fluorescence, phosphorescence, and
excitation spectra of a fraction ( 0 - 5 )
trapped from the first group of peaks in
the chromatogram are shown in Figure
6. All other fractions in the series D-1
through D-6 yielded spectra similar to
those in Figure 6 which are typical of
indoles. Similarly, all fractions in the
series 0 - 7 through 0 - 1 2 produced

INFRARED SPECTRAL DATA

Some pyrrolic N-H

and aromatic
N-H groups are present.
Methyl substitution predominates.
Each GC fraction contains numerous
isomers or compounds.
The pyridines (the major type in A-1
through A-8) are di- or trisubstituted.
The quinolines (the major type in
A-9 through A-11) are disubstituted

700

500

600

400
WAVELENGTH

Figure 7.

spectra typical of carbazoles as shown by

the luminescence spectra in Figure 7 .
Excitation near 400 mp produced broad
phosphorescence centered near 500 mp
(see Figures 6 and 7). For comparison,
spectra of some reference compounds
are included in the figures. Mean
lifetimes for phosphorescence provide
confirming evidence for the identity of
the indole and carbazole series (see
Table 111).
Although the presence of carbazoles
was easily discerned, the identity of the
indole series was less obvious from the
ultraviolet spectra. Some broad absorption near 360 mp typical of quinoidal
structures was observed.
SUMMARY OF ELECTRONIC SPECTRAL DATA

Fractions D-1 through 0 - 7 have

absorption, excitation, and luminescence
spectra typical of indoles.
Fractions 0 - 8 through 0 - 1 2 are
mainly carbazoles.
Presence of quinone structures explains broad absorption near 350 mp and
phosphorescence near 500 mp (the
fractions with these characteristics also
showed carbonyl absorption near 1700

300

2200

(mp)

Fluorescence, phosphorescence, and excitation spectra of trapped fraction D-8

A
6

Excitation spectrum (for fluorescence)

Fluorescence spectrum (excitation at 280 mp)
C Phosphorescence spectrum (excitation a t 280 mfi)
D Phosphorescence spectrum (excitation at 390 m p )
VOL. 38, NO. 1, JANUARY 1966

25

em.-' in the infrared region as discussed below).

The number of substituent carbon
atoms for the indole series (C,HZ,-~N)
and the carbazole series (C,H2,-15N)
from mass spectra (see Table IV) are

Table 111. Lifetime for Phosphorescence

of the Compounds in Trapped Fractions from Isolated Fraction D, Indoles
and Carbazoles

Mean lifetime
(sec.)
4.8
5.0
5.4
5.4
5.4
5.4
5.5
5.6
5.6
5.6
5.6
5.6

Fraction
NO.
D-1
0-2

0-3
0-4
0-5
D-6
0-7
D-8
D-9
D-10
0-11

0-12
Reference compounds
Tetrahydrocarbazole
4.8"
1,2-Dimethylindole
5.2"
Carbazole
3.7
3-Methylindole
4.8
a These values were
determined in
EPA. All others were determined in PH.

Table IV.

Fraction

( x 103)

0-2

1.06
0.63

0-3

1.34

0-4

0.94

0-5

0.63

0-6

1.28

0-7

1.32

9
11b

9
11b

9
9
llb

9
116

0.92

15
9
11*

15

26

D-9

1.89

15

D-10

0.85

15

0-11

0.75

15

0-12

0.10

15

The former type usually shows only one

carbonyl band around 1700 ern.-'
whereas in isatin the C=O frequencies
are 1772-1755 cm.-l (2-position) and
1750-1734 em.-' (3-position). Such
structures could have been derived from
introduction of two hydroxy groups
during the catalytic cracking reaction

"Identification' '
C1-Indoles
CTIndoles
Pyrroloquinone
CrIndoles
C1-Pyrroloquinone
CrIndole
CpPyrroloquinones
Cs-Indoles
(fragment ion)
CrPyrroloquinones
(&-Indoles
(fragment ion)
C3-Pyrroloquinone
C6-Indoles
(fragment ion)
C3-Pyrroloquinone
Carbazole
Cs-Indoles
CrPyrroloquinones
C1-Carbazole
(fragment ion)
C1-Carbazole
(fragment ion)
CzCarbazoles
(fragment ion)
CTCarbazoles
(fragment ion)
C3-Carbazoles)
(fragment ion)

Because one proton (weakly acidic) is easily ionized the parent-1 ion is listed.
For the series C,H,, ,OnN.

ANALYTICAL CHEMISTRY

followed by mild oxidation to the quinone. Direct air oxidation may be a

more probable mechanism, however.
Infrared spectra confirmed the presence of pyrrolic N-H groups (band a t
3485 em.-' in Figure 8) and carbazole
structures (bands a t 1230 and 1315
em.-'). These spectra also provided
some information concerning the positions of substitution. Many of the
fractions contained several compounds
or isomers, making interpretation difficult. Salient features of the infrared
spectra of these fractions are depicted
in Figure 8. The expected behavior in
the C-H out-of-plane bending region
is summarized below (11) :

Substituent position
3- but not 22- but not 3none at 2,3none at 4,5,6,74- or 7- (Le,, three adjacent

Data for C,Hz, ,N series

Largest0
Relative
2
mle
intensity
9
130
100
9
144
39
11b
147
7

116

a
b

Summary of Mass Spectral Data for the Indole-Carbazole Fractions

Per cent
of original
sample

D-1

D-8

indicated on the chromatogram in

Figure 5. Only one other significant
series appeared in the mass spectra.
This corresponded to C,Hzn-l102N and
was most concentrated in fractions 0-4
through 0-7. These fractions also
showed aromatic carbonyl absorption
a t 1700 em.-', were brownish yellow in
color, and had broad absorption around
360 mp typical of quinoidal structures.
Such carbonyl-containing structures
might also be expected to phosphoresce
by exciting radiation as high as 400 mp.
The most probable structure for these
compounds based on infrared absorption
is

Frequency
(ern.-')

810-760
785-770
725-7 10
ca. 750

H's)
790-770
5- or 6- (Le., two adjacent H's) 830-800
disubstitution of carbocyclic

ring

830-800

Fractions D-1 through 0-3 show little

substitution on the heterocyclic ring
but have bands in the 785 to 810 cm.-'
region attributed to monosubstitution
on the carbocyclic ring. In fractions
0-4 through D-6, substitution on both
rings is indicated by a decrease in the
715 em.-' band and an increase in the
absorption around 800 cm.-l Disubstitution of the carbocyclic ring is also
possible. The infrared spectra of the
carbazole fractions were similar to
each other and provided little information about positions of substitution
except that unsubstituted carbocyclic
rings were evident.
The C-H stretching region of the
spectrum indicated that most of the
substitution was in the form of methyl
groups. This was manifested as three
bands a t 2920, 2940, and 2965 cm.-l.
Some ethyl (or propyl) groups are indicated in fractions 0-4 through D-6 as
indicated by increased intensity of the
2965 em.-' band. However, even for
these fractions, methyl groups attached
directly to aromatic nuclei seem to
predominate. The relative intensity of
the aromatic C-H
absorption (between 3000 and 3100 cm.-l) compared
to the aliphatic absorption (between
2800 and 3000 em.-') provides a measure of the degree of substitution. For
example, fraction D-1 shows a small
degree of substitution (probably monosubstitution, whereas D-5 shows that
perhaps half or more of the aromatic
protons have been substituted (Figure
8). Figure 8 also clearly shows intense
N-H stretching bands for indoles or
carbazoles a t 3485 cm.-l. It is unlikely that N-alkyl substituents are
present to any significant extent.

D- 5

! ' I
io0

3400

3200

I
I
3000

I
281

1
*

FREQUENCY (CH. -1)

Figure 8. Infrared spectra of some trapped fractions from fraction

D showing the N-H and C-H stretching region and the C-H outof-plane bending region

SUMMARY O F INFRARED SPECTRAL DATA

Other Acidic Compounds (Fraction

Acidification of the hydrolyzed
sodium aminoethoxide extract yielded
a significant fraction having a dark
color and strong phenol-like odor.
The occurrence of this fraction was
surprising as a rather specific alkyl
phenol fraction (fraction C) having a
characteristic phenolic odor was first
isolated by extracting with aqueous
sodium hydroxide.
This fraction was too high-boiling to
be separated b y gas chromatography.
The total fraction was characterized
spectrometrically, however. Infrared
spectra of this fraction showed strong
0-H
stretching (3610 cm.-I) and
C-0 stretching (1200
bands
typical of phenolic compounds. A
band a t 3480 cm.-' typical of N-H
stretching in indoles or carbazoles was
also resolved in dilute carbon disulfide
solution (under which condition hydrcgen bonding was minimized). The
behavior of the ultraviolet spectra
before and after adding a base was
typical of the presence of phenolic
structures. Fluorescence and phosphorescence spectra of this fraction
were typical of carbazoles. Mass spectral data confirmed the presence of
simple alkyl phenols as the major
species. A series of compounds identified as alkyl carbazoles was observed
in smaller amount. Also found in
minor concentrations were other mass
series represented by C,H2,-130N,
C,HZ,-~SON, and C,H2,-11N formulae.
On the basis of the spectral evidence
and the weakly acidic nature of this

G).

All fractions show strong pyrrolic

N-H absorption a t 3485 em.-'
Fractions D-1 through 0-7 are indoles.
Fractions 0-8 through 0-12 are
mostly carbazoles with some indoles
carried into D-8 and D-9.
Substituents are mostly methyl
groups except for cp-and Cpindoles
which have longer alkyl groups.
The substituted indoles have alkyl
groups on both the heterocyclic and
carbocyclic rings.
Unsubstituted carbocyclic rings are
evident in the carbazole fractions.
The sharp GC peak in fraction D-1
does not correspond to the retention
time of 2-methylindole. Its infrared
spectrum does not agree with that of
reference spectra for either 2- or 3methylindole. Therefore, this peak is
most likely 4- or 7- methylindole (or
both).
In the carbazole series, the first GC
peak is quite broad and undoubtedly
contains a t least two compounds. This
peak appears between the retention
times for tetrahydrocarbazole and carbazole. The parent compound, carbazole, plus some other unidentified
related substance is probably present.
For both series of compounds, the
relative abundance is related to the
extent of substitution. From the chromatogram in Figure 5, it is obvious that
C1 and Cz indoles and carbazoles are
more abundant than the parent compounds or higher substituted species.

fraction, small concentrations of the

following classes of compounds could
also be present : hydroxyindoles, hydroxycarbazoles,
hydroxypyrroloindenes, and dihydropyrroloindenes.
This fraction was the most highly
colored of the entire set of fractions.
Partial oxidation and/or condensation
of pyrrolic compounds could explain the
almost black appearance of the fraction.
These compounds undoubtedly contribute to the color (and odor) of heating oils and other similar petroleum
fractions. Their removal would require,
on the basis of the separation scheme, a
strong nonaqueous base rather than
aqueous sodium hydroxide.
Also, it was interesting that the
aqueous raffinate (fraction F) remaining
after hydrolysis and acidification of the
sodium aminoethoxide extract was
highly colored. This color stemmed
from strong broad absorption near 340
mg which "tailed" well into the visible
region of the spectrum. The solution
showed intense broad fluorescence a t
470 mp upon excitation a t 390 mp
which could be derived from oxidation
and/or condensation products of pyrrolic compounds.
Other Observations on Extractions
Using Sodium Aminoethoxide in
Ethanolamine.
Aliphatic amides
were extracted from a 430-650' F.
distillate of Wilmington crude oil
during extraction with sodium aniinoethoxide to isolate acidic compounds.
These amides represented a very
small fraction of the material released
from the basic extract upon hydrolysis.
Phenols, indoles, and carbazoles were
first extracted into hexane. The amides,
being less soluble in pentane, were
rather selectively removed from the
hydrolyzed extract by contacting with
an ethyl ether-isopentane mixture.
Larger quantities of carboxylic acids
were isolated during this separation by
acidification of the hydrolyzed extract.
The amides were probably extracted
by virtue of the weakly acidic nature
of the N-H group. The amide structure was definitely identified by broad
absorption between 3200 and 3400 em.-',
an intense band a t 1570 cm.-', and a
band a t 1405 em.-' attributed to methylene groups adjacent to carbonyl
groups. The amides were aliphatic as
only methyl and methylene bands were
observed to the exclusion of characteristic aromatic absorption bands.
The presence of amides in crude oils
(Russian) has been reported by Bezinger
and coworkers ( 2 ) . Okuno, Latham,
and Haines (14) have reported the
presence of weakly basic compounds
(as titrated with perchloric acid in
acetic acid) which are converted to
strongly basic compounds upon reduction with lithium aluminum hydride.
This class of compounds, which would
include amides, primarily, was found
VOL. 38, NO. 1, JANUARY 1966

27

to be present in several of the crude oils

they studied (including Wilmington,
Calif., crude).
LITERATURE CITED

(1) Bender, D. F., Sawicki, E., Wilson,

R. M., Jr., ANAL. CHEM.36, 1011
(1964).
\ - - - - ,

(2) Bezinger, N. M., Abdurakhmanov,

M. A., Galpern, G. D., Petrol. Chem.
(U.S.S.R.) 1 , 13 (1962); [Neftakhimiva
1 , 23 (196l)l.
(3) Cook, G. L., Church, F. M.,J. Phys.
Chem. 61, 458 (1957).
(4) Dinneen, G. U., Cook, G. L., Jensen,
H. B., ANAL.CHEM.30,2026 (1958).
(5) Drushel, H. V., Sommers, A. L., Zbid.,
37,lO (1966).

(6) Hartung, G. K., Jewell, D. M., Anal.

Chim. Acta 26,514 (1962).
(7) Hartuna, G. K., Jewell. D. M.. Zbid..
27,219 (fi63).
(8) Heinert, D., Martell, A. E., J. Am.
Chem. SOC.81, 3933 (1959).
(9) Helm, R. V., Latham, D. R., Ferrin,
c. R., Ball, J. S., ANAL.CHEM.32, 1765
11960)
\ _ _ _ _

(10) K&r, C., Jr., Estep, P. A., Papa,

A. J., J. Am. Chem. SOC.81, 152 (1959).
(11) Katritzky, A. R., Physical Methods
in Heterocyclic Chemistry, Vol. 11,
p. 212, Academic Press, New York, 1963.
(12) La Lau, C., Anal. Chim. Acta 22,
239 (1960).
(13) Lochte, H. L., Pittman, A. G., J.
Org. Chem. 25, 1462 (1960).
(14) Okuno, I., Latham, D. R., Haines,
W. E., ANAL.CHEM.37,54 (1965).

(15) Sauer, R. W., Melpolder, F. W.,

Brown, R. A., Ind. Eng. Chem. 44, 2606
11952).
(16)-Snyder, L. R., Buell, B. E., ANAL.
CHEM.
34,689 (1962).
(17) Snyder, L. R., Buell, B. E., Zbid.,
36,767 (1964).
(18) Thompson, C. J., Coleman, H. J.,
Ward, C. C., Rall, H. T., Ibid., 34, 151
(1962).

RECEIVED
for review February 18, 1965.
Accepted October 7, 1965. Division of
Petroleum Chemistry, 149th Meeting
ACS, Detroit, April 1965. The authors
thank Esso Research Laboratories, Humble Oil & Refining Co., for permission to
publish this material.

Simultaneous Determination of Lead Alkyls and Halide

Scavengers in Gasoline by Gas Chromatography
with Flame Ionization Detection
NESTOR L. SOULAGES
laboratorio Petrotdcnico, Yacimientos Petroliferos Fiscales, Florencio Varela, Repliblica Argentina

b A gas chromatographic method

with flame ionization detection for
the determination of lead alkyls and
scavengers in gasolines is described.
These compounds are separated in a
partition column and hydrogenated
using a nickel catalyst, and the resulting methane and/or ethane is separated from gasoline hydrocarbons on
an adsorption column. The equipment
employed and its application to the
continuous analysis of leaded gasolines
are discussed. The reliability of results and the lack of interferences are
verified with both laboratory-prepared
and commercial samples.

steady increase in the use of mixtures of methyl-ethyl lead alkyls as

antiknock agents requires rapid and safe
analytical methods. Total lead content
in a gasoline as well as lead distribution
should be determinable.
Since the introduction of the electron
capture detector by Lovelock and
Zlatkis (6), its possibilities have been
used (1,4, 5 ) in spite of the lack of reliable response and difficulty in avoiding
unknown interferences.
Parker, Hudson, and Smith (7,8) have
made chromatographic fractions react
with iodine solutions prior to the colorimetric evaluation of lead. Absence of
interferences is outbalanced by the
elaborate and time-consuming technique.
Because of the complexity of a gasoline, it is impossible to think of a selecHE

28

ANALYTICAL CHEMISTRY

tive separation efficient enough to

permit collection of the lead alkyls and
scavengers isolated from all the hydrocarbons present. With electron capture
detection or colorimetric detection, sensitive only to lead, hydrocarbon interference is eliminated by the detector and
the chromatographic column has only to
separate the lead alkyls and scavengers
from each other.
The present method uses the catalytic
hydrogenolysis of the lead and halogenated compounds and measurement
of the resulting methane and ethane,
which are easily separated from gasoline
hydrocarbons.
Hydrogenation catalysts and their
chromatographic application have been
extensively studied by Thompson et al.
(9-11) and Beroza (2, 3) from a qualitative viewpoint.
EXPERIMENTAL

For a better understanding the process can be divided into three stages
which are conducted continuously:
Separation of lead alkyls and scavengers in a partition (GL) column using
polypropylene glycol 400 (PPG) as
stationary phase. In this step additives are simultaneously separated from
methane and ethane potentially present
in the gasoline, thus avoiding further
interferences.
Catalytic hydrogenolysis of lead
alkyls, leading to the formation of
methane, ethane, or both, according
to the original compound, and simultaneous hydrogenation of ethylene di-

chloride and ethylene dibromide to

ethane. Catalysis was accomplished
by passing the carrier gas, hydrogen,
over nickel a t a suitable temperature
to avoid hydrocarbon cracking. Double
bonds, both olefinic and aromatic, are
hydrogenated only under the experimental conditions.
Separation of methane and ethane
thus formed, from normal gasoline
hydrocarbons-Cs
and higher-in
a
(GS) chromatographic column packed
with activated charcoal with final detection by flame ionization.
Figure 1 shows schematically these
three stages of the process.
Equipment. Figure 2 is a diagram
of the equipment showing intermediate connections between the various
parts and the path of the carrier gas.
The oven contains the preliminary
partition column, a ten-way valve, and
two adsorption columns. Nickel catalyst remains outside, as its working
temperature is different from that
within the oven.
Broken lines indicate thermostated
zones. Variable transformers allow for
individual control and adjustment of
temperature within the oven, the catalyst container, and injector. The oven
heating resistance is isolated to avoid
explosion hazards due to eventual
hydrogen loss.
The flame ionization detector and
circuits are of conventional design.
Hydrogen is purified by passing it along
a charcoal column kept a t liquid nitrogen temperature to minimize background and improve base line stability.
Columns. The partition column
is prepared by filling a 20-cm. length