EPA-600/2-77-163
August 1977
CORRELATION OF OILS AND OIL PRODUCTS BY
GAS CHROMATOGRAPHY
by
0. Zafiriou, M. Blumer, J. Meyers
Woods Hole Oceanographic Institution
Woods Hole, Massachusetts 02543
and
D. Stainken
U.S. Environmental Protection Agency
Edison, New Jersey 08817
Grant No. 15080HEC
Project Officer
M. Gruenfeld
Oil and Hazardous Materials Spills Branch
Industrial Environmental Research Laboratory - Cincinnati
Edison, New Jersey 08817
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH & DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Industrial Environmental Research
Laboratory, U.S. Environmental Protection Agency and approved for publi-
cation. Approval does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
ii
-------
FOREWORD
When energy and material resources are extracted, processed, converted,
and used, the related pollutional impacts on our environment and even on
our health often require that new and increasingly more efficient pollution
control methods be used. The Industrial Environmental Research Laboratory -
Cincinnati (lERL-Ci) assists in developing and demonstrating new and improved
methodologies that will meet these needs both efficiently and economically.
This report is a product of the above efforts. It establishes a method-
ology for correlating (fingerprinting) weathered and unweathered oils by gas
chromatography. A routine method of sampling, sample extraction and analysis,
and interpretation is reported. The gas chromatographic method established
in this report can be utilized as a rapid, efficient and routine method for
correlating environmentally weathered petroleum oil pollutants with source
oil samples. This report will be of interest to those individuals involved
in routine monitoring of the environment, damage assessment, law enforce-
ment, and industrial wastewater research. Further information about this
report is available from the Oil and Hazardous Materials Spills Branch,
lERL-Ci, Edison, NJ 08817.
David G. Stephan
Director
Industrial Environmental Research Laboratory
Cincinnati
iii
-------
ABSTRACT
A gas chromatographic method is presented for identifying the discharge
source of petroleum oil pollutants found as slicks and shoreline residues,
The method is used to match environmentally altered oils with unweathered
source samples, a technique known as "fingerprinting". Analyses of arti-
ficially aged oils and potential spill sources found in Greater New York
Harbor and Portland, Maine, indicated a high rate of success for the method
in realistic situations.
The method was demonstrated to be suitable for routine use with weath-
ered and unweathered samples, and for monitoring levels of hydrocarbons in
organisms and sediments. The method can be modified to study the fate and
effects of lower levels of petroleum hydrocarbons.
This report was submitted in fulfillment of Grant No. 15080HEC by
Woods Hole Oceanographic Institution under the sponsorship of the U. S.
Environmental Protection Agency. This report covers the period January 1971
to January 1972 and work was completed as of October 1972.
iv
-------
CONTENTS
Foreword ill
Abstract ±v
Figures vii
Tables viii
Abbreviations and Symbols ix
Acknowledgement x
1. Introduction 1
2. Conclusions 3
3. Recommendations 4
4. Gas Chromatography of Oils 15
5. Sampling Procedures 15
Source oils 15
Oil in water 15
Oil in sediments and organisms 16
6. Sample Preparation 18
Oil-rich samples 18
Oily water, sand straw etc 19
Sediments and Organisms 20
Sample Extraction 20
Column Chromatography 21
Removal of Sulfur from sediment samples 22
Preparation and use of standard reference samples 23
Standard No. 2 fuel oil 23
Standard sediment sample 23
Standard organism sample 23
Artificial weathering 24
7. Sample Analysis 25
Preparation of gas chromatograph and column 25
Analyses and daily operation 28
Maintenance procedures and standards 29
Chromatogram measurement and data tabulation 30
8. Interpretation of Data 34
9. Results and Discussion 38
Ancillary techniques 40
Practicability 42
References 44
Bibliography 46
Appendices 48
A. Petroleum Chemistry 48
B. Petroleum geochemistry 53
v
-------
C. Petroleum weathering 54
D. Sample Chromatograms 57
E. Intercalibration analyses of recently biosyntheslzed 74
hydrocarbons and petroleum hydrocarbons in marine lipids ....
F. Procedure for cleaning Varian capillary injector 76
G. Source-spill correlation data for EPA test series 77
vi
-------
FIGURES
Number
1 Glass-lined capillary injector 6
2 Relation of peak area ratios to oil injection ratios 14
3 Chromatogram showing definition of terms 31
4 Chromatogram of Bachaquero crude oil 51
5 A geochemical source of pristane and phytane 55
Yii
-------
TABLES
Number
1 Long-term Variability of Analytical Ratios of the Prepared
No. 2 Fuel Oil Standard
2 Variability of Consecutive Analyses of Crude Oils for Several
SCOT OV-101 Columns 10
3 Intercolumn Comparison of Ratio Means
(Expressed as Absolute Values of Ratios) 13
4 Conditioning Sequence of SCOT OV-101 Columns 25
5 GC Operating Conditions for Oil Analysis 28
6 Correlation of 17 Source Oils and 35 Simulated Spills
by Gas Chromatography 38
7 Percent Success of Gas Chromatographic Correlation
of "Spills" and "Sources" 39
8 Ability of Gas Chromatography to Distinguish Among 30 Oils
and Oil Products from Greater New York Harbor 41
D-l Chromatography Column, Injection Number and Miscellaneous
Parameters used to Obtain Sample Chromatograms of Appendix D . . 57
D-2 Data for Oil and Oil Products Obtained from the Sample
Chromatograms of Appendix D 59
E-l Intercalibration of Hydrocarbon Analyses
(IDOE-5):ug hydrocarbons/g cod liver lipid 75
G-l Source-Spill Correlation Data for EPA Test Series 77
viii
-------
ABBREVIATIONS AND SYMBOLS
BG background
GC gas chromatograph
MRT methane retention time
MS mass spectrometer
mv millivolt
MW molecular weight
Phy phytane
Pris pristane
RSD relative standard deviation
RT retention time
SCOT support coated open tubular column
SD standard deviation
Ug microgram
yl microliter
IX
-------
ACKNOWLEDGMENTS
This report has been extensively revised and edited by Dr. Dennis M.
Stainken from the original Woods Hole Oceanographic Institution Technical
Report 72-55, July 1972. Explicit recommendations and conclusions have
been added. For their valuable assistance, the authors thank Richard T.
Dewling, Project Director, Michael Gruenfeld and Frank Freestone of the En-
vironmental Protection Agency, Lt. Foley of the United States Coast Guard,
Gardiner Hunt of the Marine Environmental Protection Commission, and Dr. John
Farrington of the Woods Hole Oceanographic Institution, author of Appendix F.
-------
SECTION 1
INTRODUCTION
Petroleum oil pollutants frequently occur in the environment. These
pollutants are extremely complex mixtures containing many classes of chemical
compounds. Hydrocarbons are a major class of petroleum constituents and in-
clude the subclasses normal, branched, cyclic, saturated and aromatic hydro-
carbons. The subclass of normal hydrocarbons contains 10 to 60 individual
members, but most other oil subclasses have many more representatives. A
similar degree of complexity is found in other classes of compounds that
occur in petroleum oil heavy metals, heteroaromatics, carboxylic acids,
and mercaptans. A further discussion of petroleum chemistry and geochemistry
are provided in Appendices A and B.
A fundamental aspect of oil chemistry is the complexity of composition
both among and within categories. All feasible analyses of oils are partial,
and few actually account for more than a small percent of the total weight
of material analyzed. Experimental error introduces further inaccuracies in
the levels of components analyzed. The incompleteness of the analysis may
mean that the components in which two oils differ most are not determined.
The complexity of composition also forms a basis for chemically iden-
tifying petroleum oils. All oil correlation is fundamentally the comparison
of two or more oil samples (weathered and unweathered). Weathered and un-
weathered source oils are analyzed by gas chromatography, and the chromato-
grams are compared visually and by means of specific peak-height ratios.
Two samples of oils with identical chemical compositions must have identical
histories. This matching of a source oil with an environmentally altered
sample oil is termed "fingerprinting". However, very similar oils may not
be distinguishable. The strongest statement that can be made is that the
oils compared are definitely different or that they are not distinguishable
by the method used.
The study objective was to develop a method for identifying the dis-
charge source of a petroleum oil pollutant, i.e. a gas chromatographic method
for matching a weathered oil with an unweathered source sample of the oil.
Two test series were used.
In the first tests, an assortment of common oils and oil products were
weathered under varying conditions and exposure periods by the U.S. Environ-
mental Protection Agency (EPA), Edison, N.J. Several spill-control chemicals
(6% by weight) were also added to some samples. Of 17 source oils, 16 were
sent to Woods Hole along with 35 simulated spill samples. Samples were coded
and identified only as potential sources or simulated spills. Appendix G
1
-------
contains the list of samples. For each spill sample, all 16 source samples
were considered. Over 10^^ possible sets of correlations existed.
A second test of the method was made by obtaining representative sam-
ples of oils occurring in two areas and analyzing them. Samples were ob-
tained from the cargoes of 22 tankers in the Port of Portland, Maine. The
U.S. Coast Guard provided cargo samples of 42 ships, barges, and dock facil-
ities from the New York Harbor area.
The method was further tested by establishing and analyzing standard
reference samples. The reference standards prepared for these analyses were
a No. 2 fuel oil, cod liver oil spiked with the No. 2 fuel oil, and sediments
contaminated with the No. 2 fuel oil.
-------
SECTION 2
CONCLUSIONS
Gas chromatography (GC) can be used to identify the discharge sources
of petroleum oil pollutants. Identification is accomplished by a process
termed "fingerprinting", in which an environmentally weathered oil is matched
with unweathered samples of the possible source oils.
Fingerprinting of weathered and unweathered petroleum oils is accom-
plished by comparing the ratios of peak heights of n-C-^, pristane, n-C^g,
and phytane above a traced baseline. Fingerprinting is further substantiated
by comparing the ratios of unresolved substances, which may be obtained by
drawing a tangent line across the base of the phytane/n-C^g and pristane/
n~^17 Peaks. This ratio, which is based on the vertical line drawn from the
baseline to the tangent line through the phytane peak, is less subject to
alteration by evaporative weathering or by bacterial action on the oil than
is the previously recommended n-C-^y background ratio. Additional comparison
is achieved by direct visual inspection of the gas chromatograms of oils.
Normal hydrocarbons derived from plants, organisms, clean recent sed-
iments, and some shales exhibit a strong predominance of odd-numbered over
even-numbered compounds. Petroleum-derived material exhibits odd/even ratios
near 1:1 and a complex unresolved mixture. This difference can help deter-
mine the origin of samples.
The methodology described here may also be used for correlating un-
weathered source oils with environmentally weathered petroleum oils extracted
from contaminated organisms and sediments.
-------
SECTION 3
RECOMMENDATIONS
The gas chromatographic method described within this report is recom-
mended as the routine method for identifying the discharge source of petrol-
eum oil pollutants.
A data file of chromatograms and calculated ratios of crude oils that
are produced in various geographic areas should be established. Gas chro-
matographic analyses of sediments and organisms would be useful for deter-
mining baseline data.
Standardized methods of sampling and sample preparation procedures
should be established.
Gas chromatographic analyses and identification of oil and oil products
can be further enhanced by use of mass spectrometry and fluorescence spec-
troscopy.
-------
SECTION 4
GAS CHROMATOGRAPHY OF OILS
Most analytical difficulties in gas chromatography arise because the
interaction between the resolving material in the GC column and the sample is
very strong. The sample passes through the gas chromatographic column and is
likely to modify it irreversibly to some extent.
Oils present particularly difficult samples because of their wide
boiling range and polar as well as nonpolar materials content. Relatively
high resolution is required to obtain necessary separations. The higher res-
olution requirements increase the difficulty of preventing sample modifica-
tion of the column.
Two basic approaches can be used to resolve this problem. Samples can
be pre-fractionated into simpler and cleaner materials for analysis or mini-
mally treated samples can be analyzed in a GC system optimized to provide
reliable operation.
Analysis of minimally treated samples requires less operator time and
skill and allows the factors that introduce variability to be more tightly
specified. This approach was used here. Pre-fractionation of samples can
yield more information, however, especially when performed by experienced
personnel. But this procedure can increase variability if improperly used,
and it is therefore not suited for large-scale programs or routine identi-
fications of petroleum oils.
In most analytical situations, fresh oils suspected as spills sources
must be compared with environmentally altered samples. The weathered and
unweathered oils have different histories and are therefore expected to have
different compositions. To trace the source of the environmental sample, one
must be able to specify the differences between source and spill that can
reasonably be expected to arise from weathering. Any significant differences
not attributable to weathering establish that the samples are different.
Fingerprint analyses can theoretically correlate all identical oils.
Errors and incompleteness of analysis limit this capability, and weathering
introduces further differences between an environmentally exposed oil and its
source. Chemical comparison of oil samples leads to a unique source-sample
correlation only if the spill and all possible sources in the area are anal-
yzed and all but one possible source shows compositional differences from the
spill sample that cannot arise by weathering alone.
-------
GLASS LINER IN INLET TUBE
MADE SLIGHTLY
SADDLE-SHAPED
2mm
r ^-^ ,
\ . __/~r ., . _ . ., , . _ . , -'
V .-^^7 i
x t
^ Fi^k mm Ifc
EXACT DIMENSIONS
TO FIT
GLASS-LINED INJECTOR ASSEMBLY
SEPTUM
COLLAR NUT
CARRIER GAS
INLET
TO SPLITTER
INJECTOR BODY
Figure 1. Glass-lined Capillary Injector
6
-------
For fingerprinting, the method described here accepts oil samples as
dilute solutions in purified carbon disulfide. These solutions are easily
prepared from oils and oil-rich water and sediments. Spill-source corre-
lation of oils extracted from dilute water samples, sediments, and organ-
isms requires: a) more elaborate sample preparation procedures, b) analysis
of comparable unpolluted samples for background determination, and c) careful
interpretation.
Carbon disulfide solutions of oil sample are introduced by syringe into
the capillary injection port of the instrument. Since the samples often
have a nonvolatile residue, we have devised a removable glass liner for cap-
illary injection ports much like those conventionally utilized with non-
splitting inlets (Figure 1). The method utilizes the trapping property of a
cold column to eliminate the necessity of using the conventional high in-
jection port temperatures (300 to 350 C) often used for oils. This elimi-
nates column-degrading residue pyrolysis and septum bleed, and it repro-
ducibly introduces the sample onto the column. Both pyrolysis and bleed
load the column with high-boiling materials that shorten column lifetime and
increase variability. An injector temperature of 200 C is utilized, and the
early-eluting compounds are flashed into the column.
The injection sample is split to an optimal size and the sample is fed
to the support-coated open tubular (SCOT) column coated with the nonpolar
liquid silicone OV-101. This choice of column is close to the optimum for
the applications discussed. Polar-packed columns can separate pristane/C.. 7,
and phytane/C-.o well, but they do not have the temperature stability to
handle higher-boiling materials. Nonpolar-packed columns do not separate the
pristane/C,7 and phytane/C1R pairs well unless they are long and are operated
at high temperatures.
Capillary columns with various coatings render excellent separations of
the peaks of interest, but they are sensitive to deterioration or destruction
by addition of high-boiling materials. They have a short lifetime unless
samples are pre-treated to eliminate residues.
SCOT capillary columns offer a compromise between packed-column dur-
ability, and wall-coated capillary column resolution and low retention. The
length and openness are capillary-like, but the support-coated wall, with its
much higher load of stationary phase, increases capacity and durability. The
flow resistance is low and reproducible from column to column.
The silicone OV-101 coating chosen for the SCOT column has several ad-
vantages. OV-101 provides good separation of both the C17/pristane and the
C..o/phytane pairs. The separation is complete enough that changes in res-
olution (peak sharpness) hardly affect most ratios, facilitating comparison
and correlation. The use of helium as a carrier gas is essential for good
resolution performance with SCOT columns. The second advantage of the OV-
101 substrate is its high thermal stability. The column is always operated
at least 75 C below its maximum recommended temperature, prolonging column
life and minimizing bleed of silicones into the detector and subsequent
fouling.
-------
SCOT columns maintain superior performance when in continuous service.
Older columns yield low resolution chromatograms after a dormant period,
but repeated analysis of No. 2 fuel oil provides a return to good resolution,
which is maintained until a new period of disuse. The permissible shutdown
period and reconditioning required eventually become excessive, though the
chromatograms produced under optimal conditions may be of good quality.
SCOT columns of Apiezon L and of Dexsil 300 (Chromatogram 52) were
tested, but neither separated n-C from phytane well. Further improvement
over SCOT OV-101 column performance would require capillary columns and
preliminary separations to protect them.
Special innovations in detection (flame ionization is used) or recording
were unnecessary, though these components must function properly. The tem-
perature programmer of the GC must be reproducible and calibrated.
To ensure that the analytical system is performing correctly, the
analyst frequently runs blank determinations and chromatograms of a standard
oil sample. A No. 2 fuel oil stabilized against chemical decomposition by
removal of polar impurities has been chosen. Frequent analysis of the stan-
dard is an extremely important part of the procedure. It permits the
analyst to detect and correct for small variations in column performance
and to detect any increase in variability of results that suggests some
fault of the equipment or technique.
Variability is detected by comparing peak height ratios of specific
(i.e., pris/phy; C17/pris; C-g/phy; C ^ i 'C ^ C^ /Background;
valley) by means of relative standard deviations (coefficient
17-17_ .
of variation) . The relative standard deviation (RSD) of each peak-height
ratio is determined as the ratio of the standard deviation to the mean.
RSD of a peak height ratio = S
where S = the standard deviation of the given peak-height ratios
X = the mean of the peak-height ratios
The RSD is frequently expressed as a percentage (i.e., 0.258 RSD = 25.8%).
The variability of a given ratio determined on the No. 2 fuel oil is appli-
cable to crude oil analyses also. Estimating the variability of ratios is
important in interpreting the significance of proposed correlations of oil
fingerprinting.
Table 1 presents the RSD of several analytical ratios for a No. 2 fuel
oil standard. The analyses span considerable use periods and three dif-
ferent columns in three different instruments (of the same make and model).
-------
en
H
P-I cu
1 rH
1
Q
"^
H
C/3
^j
M
o
W
PM
CM
O
Q
2
^
P-I
w
pi
P-I
w
tn
H
^
o
en
o
M
H
PH
1
^
O
I-H
H
cO
CO
O
H
4-1
rl
HH
0
CO
C
O
4-1
cfl
, J
*rn
0)
nd
Td
^_i
CO
13
C
cfl
4-1
CO
cu
H
4-1
Cfl
CU
c£J
r*»*
r 1
u
PQ
r^«
rH
U
00
u
!^
u
K**»
Lf~{
P-I
00
rH
O
en
H
t-l
P-I
rH
^
^">
»^
CM
CO
H
l-l
P-i
° .
0
^
C
o
H
4-1
O
cu
i)
C
M
w
t3
3
UH
O
o
iz
, )
CO
4J
O
H
pi
e
3
i-H
O
U
1 CN
< CN vD
v£5 vD vO
m oo in
CN CN rH
m o ON
oo oo m
CN rH r-l
rH 0 0
rH CO VO
CO CM CM
m
JM| J^
3 .- cu
60 13 4J
H 0) C
pm l-i tH
01
C ^a cu
34-1
0 C
cu oo
C CO C3
H -H -H
CU C 3
13 O 13
H
CO 4-1 13
CO O CU
CU rl
*O *r-) ^|
CU C 3
l-i -HO
3 0
CO CU O
Cfl rH
0) O, CO
6 S 13
Cfl rl
CO CO Cfl
O 13
H J£ C
4-> CJ Cfl
CO Cfl 4-1
p> W co
CO ,0
-------
The relative standard deviations are largely in the range of 1.6% to 3.5%.
Deterioration of the columns appears to be a minor factor. Note that only
three of the first four ratios are independent, as four independent quan-
tities are ratioed. Variability related to fractionation and temperature
effects should make the largest error contribution to the C-
and
'17/18
pristane/phytane ratios. Shifts in elution temperatures or resolution
changes affect primarily the C.. 7/pristane and Cis/phytane ratios. The
C.j-/background ratios should be particularly sensitive to resolution changes.
Table 2 lists the RSD's obtained with three columns for several sam-
ples when analyzed repeatedly in a short time period. With a few exceptions
(caused by unusual sample type or evident instrument malfunction), these
RSD values are about half those found over column lifetimes. The better
performance of the second and third columns is clear. The high variability
of ratios obtained from column 219 suggested vaporization-related problems.
Cleaning the injector led to an immediate decrease in the variability of
these ratios.
TABLE 2. VARIABILITY OF CONSECUTIVE ANALYSES OF CRUDE OILS FOR
SEVERAL SCOT OV-101 COLUMNS
% Relative standard deviations
o
of ratios
Column No. of Crude oil
runs
Pris/phy C17/pris C^/phy C^/C^
216°
218
218
219
6
6
6
6
TJ It
Agha
Agha
Agha
. (crude)
Jari
Jari
Jari
3.
2.
2.
5.
94
20
10
83d
3.00
0.70
0.82
1.30
2.
1.
0.
1.
65
07
91
72
4
2
1
6
.16
.87
.75
.iod
9.59
3.41
1.85
2.68
a Ratios measured as defined in Figure 3 and related discussion.
b Columns used had been utilized previously for 80 to 160 analyses.
c After period of disuse; variability may be caused in part by "atrophy
effect."
d See text. Variability decreased after injection port cleaning.
The properly operating system yields RSD values on the order of 0.7% to
3% over short time periods, and less than twice this level over longer time
periods. Failure to achieve this level of reproducibility suggests de-
ficiencies in columns, instrumentation, or technique.
10
-------
Table 3 compares the absolute values of ratios obtained for the same
sample on different instrument/column combinations. The three tested col-
umns appear similar. For several ratios, the differences among columns are
not over one standard deviation for the repeated analyses of the same sample
at the same time. The greatest ratio differences are those affected by
resolving power. Column No. 216 was definitely somewhat less efficient.
The smaller ratio differences are often unimportant and larger ones are
compensated by a correction factor (discussed below) to make results
comparable.
A correction factor relating two columns, A and B may be derived by
analyzing the same sample on each column (this need not be done at the same
time). The sample used should have a rough similarity to the other samples
involved; i.e. the ratios for which correction factors are desired should
be within + 20%.
Correction Factor (for samples run on column B and compared to samples run
on column A):
F = ratio determined on column A
ratio determined on column B
A comparison of peak height ratios for the same pair of peaks of other
samples can be made by multiplying the ratio determined on column B by the
correction factor to obtain results comparable with other samples analyzed
on column A. Similarly, 1/F can be used to convert results on column A to
those on column B.
RSD and ratio comparisons illustrate the degree of reproducibility ob-
tainable. Some of the variability in these ratios has been traced to sample
size factors that can be controlled. Comparisons of similar-sized samples
have smaller ratio variabilities. The effect of sample size on various
signal intensity ratios is shown in Figure 2. The results were obtained in
a study of sample size vs. ratio values performed on column 219 using a
prepared standard No. 2 fuel oil. The x axis is a dimensionless quantity.
It is the ratio of the two sample volumes in units of microliters: (ex.
injection 1 - yl oil in CS_/injection 2 - pi oil in CS_ = unitless number).
The injection size was constant. Figure 2 illustrates that the signal in-
tensity ratios are little affected (except for n-C17/BG) by different sam-
ple volumes. Such plots may differ from column to column and, to a small
extent, with sample type.
Records of column behavior with a standard sample should be maintained.
A standard should be analyzed several times consecutively to check procedure.
Increases in variability or change in absolute ratios indicate necessary
corrective action. The relative standard deviations obtained from one sam-
ple type seem to be applicable to other samples. Estimates of the signi-
ficance levels of observed ratio values can be made for spill samples anal-
yzed once. For example, if pairwise comparison of data were necessary for
all samples, the data in Table 7 (Section 10) would require over 1,000 GC
analyses, expending at least one column in the process. In fact, duplicate
11
-------
analyses of all samples and accompanying No. 2 fuel oil standards and
blanks enabled us to make most comparisons with 120 analyses. Only a few
very similar pairs were then re-analyzed in an "ABA.B" format. This pro-
cedure yielded the data in Table 7.
12
-------
^ s
CO
O
H
2
Fn
0
CO
w
J3
>
W
H
rJ
0
CO
pa
CO
O
W
CO
CO
w
Pi
s
5-
CO
§
si
*rj
O
[. I
H
Si
Fn
O
J^
o
c/>
M
^3
p -
*rj
o
o
g
g
rJ
O
o
&
w
H
M
en
W
PQ
<3^
H
O
pq
--^
^
o
cfl 00
CO rH
0 U
H --.
4_j r^
Cfl rH
to O
4-1 >-,
H ,£3
co a.
c ~^
0) 00
4-1 r-H
a o
H
rH Cfl
CB -H
C to
00 ft
co r>.
0
^"1
t(^]
a
CO
H
r<
PM
OJ
rH
ft
e
cfl
CO
»rj
0)
60
cfl
0)
^
cfl
CO
C
3
M
O
iz
o
J^
d
e
3
rH
O
O
vO <± <>4 \O C^
CO CO "^" i-O *^f
*
CT> **o in in
H rH rH rH rH
t i i I CTN r^ cn
oo
rH rH i-H O 0
H rH rH
H -H -H
O O O
rH rH i-H
0) 0) 0) -H -H
3 3 3 !-i to
[V | pr [ pf | [rj (rj
^ ->
CM CM CM
Cfl Cfl
rC rC
O O O 60 60
rH rH m VO VO
rH rH r-l
CJ O TJ T3 T3
OO CTi ^O OO ON
r-l rH 1 1 rH rH
CM CM CM CM CM
.
4J
X
0)
4-1
a
H
T)
0>
CO
CO
O
CO
H
T3
T3
C CO
cfl C
0
en *H
4J
0) a
to o)
3 T-)
60 fi
H -H
j"T)
0
C m
H rH
T3 m
01 O
C
J *r1
rl vj
MH O
01 -H
TJ SH CO
a) oi
CO ft CO
cfl >,
60 rH
T) CU C cfl
OI rH -H d
tJ ft J-i cfl
3 S 3
co cfl tj ai
cfl CO >
0) 0) iH
g 13 -a 4J
CO cfl 0 O
O T3 01
H C CO CO
4-* Cfl Pi CJ
cfl 4J 3 O
pi co pi o
Ctf ^2 O T3
13
-------
>^
5
X
K
C/)
§
K
^
^
v^
55
4.5
4.3
4.1
3.9
3.7
1.9
1.7
1.5
13
. .
A
- /COLUMN *2I9\
- \#2 FO. STD. / A
-
A
A n-Cl7/BACKGROUND
n-Cl8/PHYTANE
A A o PRISTANE/PHYTANE
A n-Cl7/PRISTANE
D n-Cl7/n-Cl8
*
m
~ o
0
- a
_
g m
- 8
i i i i i i i i
1 1
0 20 40 60 80
OIL INJECTION RATIOS
Figure 2. Relation of Peak Area Ratios to Oil Injection Ratios,
i.e. Inj 1/lnj 2 (Microliters Oil in CS2). Injection size constant.
14
-------
SECTION 5
SAMPLING PROCEDURES
SOURCE OILS
These samples are generally oil without other admixed material and are
obtained from tanks, pipelines, holds, etc. The sampling location and
method (pipeline withdrawal, bucket sample, sample thief, etc.) should be
recorded. The sample must represent the material of interest; for example,
pipeline "T" connections, etc. may contain previously handled materials and
require flushing; some vessels carry different cargoes in different holds.
The analyst or his representative should be present at the sampling oper-
ation. A photograph of the sampling location is also highly preferable.
A few drops of sample are adequate, but an ounce to a pint is desirable.
Inhomogeneity (for example, possible layering in tanks) requires obtaining
several sub-samples from different locations. Glass containers or metal
cans with tight closures should be rinsed with distilled methanol and pen-
tane and dried. They should have tight-sealing caps of Teflon or oil-
resistant plastic. Rubber or waxed paper cap liners are not satisfactory.
Each sample should be accompanied by an adequate description, including
location, sampler, method, and precise element (tank, line, etc.) sampled.
Any spill control chemicals utilized should be sampled in this same manner.
Samples should be refrigerated if held more than a few days after col-
lection. Sealed containers nearly full of oil and devoid of air layers are
stable for months or longer. A most important point in source sampling is
to obtain samples of all possible sources in the area as soon as possible
after a spill. An oil cannot be correlated with a sample that is not avail-
able.
OIL IN WATER
Only micrograms of oil are required for analysis. However, samples
must be representative of the contaminant. Larger samples (several ml to a
pint) are more likely to be representative and also provide material for
further analyses. Substantial discharges should be sub-sampled at a variety
of locations. Similar materials should be sampled from nearby unaffected
areas to serve as controls.
So long as they do not contaminate, several methods of obtaining oil-
rich samples are acceptable. The apparatus must be free of oil-soluble
components. Electronics grade fluorocarbons are good solvents, non-
flammable and they have minimum toxicity. They can be used safely in the
15
-------
field to clean emergency sampling equipment and containers. Oil has been
collected by dip-netting, by picking up lumps, by slowly submerging wide-
mouthed jars or buckets, by scooping along the edge of "booms", by immersing
a mop in a slick and wringing it out, and by collecting oil-soaked straw or
other cleanup agents. Methods which use clean materials and obtain rela-
tively oil-rich samples are most satisfactory. Dilute samples should be
accompanied by blank samples if apparatus of uncertain reliability is used.
A major advantage of collecting concentrated oil samples is that rapid al-
teration by micro-biological growth is less likely than for dilute oil in
water, or oil soaked in nutrient-rich organic matter such as hay. If bac-
terial seeding of oil spills is ever adopted, sampling procedures will re-
quire modification to preserve samples more effectively. Samples should be
prepared for analysis or frozen as soon as possible.
Very dilute oil in water samples may be collected by using classical
methods of oceanography and hydrology. These samplings will require careful
apparatus preparation, avoidance of ship contamination (as from bilgewater
and lubricants), and hydrographic wire lubricants. Immediate sample ex-
traction or preservation to inhibit bacterial growth will also be necessary.
Such dilute samples are research problems or sophisticated monitoring tasks
and are not suitable for routine analysis.
OIL IN SEDIMENTS AND ORGANISMS
Shallow water sediments and organisms can be sampled by simple jar
collection. In deeper water, a pre-cleaned unlubricated dredge or grab
sampler may be used. Fishnets and other shipdeck equipment are often oil
soaked and unsuitable for use without cleaning and obtaining a blank sample.
The surface 2.54 - 5.08 centimeter layer of sediments obtained by grab
sampler (Van Veen type) is separated by "spooning" this layer into a pint
mason jar. The grab used should protect the sample from excessive water
washing or homogenation. Short (30.5 - 61 centimeter) sediment cores ob-
tained with an oil-free apparatus are also very useful samples. Sampling
locations should be chosen after considering the physical and biological
character of possible locations.
Organisms and sediments should be frozen as soon as possible after
collection to prevent extensive bacterial growth. Control areas or organ-
ism populations should be established and sampled in nearby similar loca-
tions to provide comparison or baseline data. Relatively large commercially
valuable organisms are often selected for analysis. Bottom-dwelling (ben-
thic) organisms that are unable to escape an area are more likely than fish
to show clear contamination. Sedentary organisms are especially suitable
for study. The same population may be repeatedly subsampled over a period
of time.
Molluscs are particularly suitable for analysis. If an "effects" study
is contemplated, sampled organisms should be selected in consultation with
biologists. Samples should be taxonomically identified before being des-
troyed by analysis, or individuals may be preserved by classical biological
methods.
16
-------
Sediments in locations not known to have been affected by major oil
spills have been found to contain high levels of oil-derived hydrocarbons
in some locations. These presumably arise from chronic discharges which may
have similar effects as those from discrete spills in some locations. Con-
sequently, it is desirable that baseline organisms and sediment samples be
taken from locations subject to water quality control management.
17
-------
SECTION 6
SAMPLE PREPARATION
The purpose of this step is to prepare a carbon disulfide solution
(ca. 5% to 10% wt/vol) of a portion of sample for GC analysis without (a)
loss of the components utilized in correlation or (b) introduction of contam-
inants. The concentration need not be known exactly. For oils and rela-
tively oil-rich samples, these solutions are easily prenared as outlined
below. More elaborate procedures and controls become necessary for progres-
sively more dilute samples admixed with larger amounts of matrix and with
naturally occurring hydrocarbons. However, when treatments to remove inter-
fering materials are required, the ability to discriminate among alternative
source oils diminishes. In these cases, the presence of oil contamination
can still be established under conditions that preclude correlation with a
specific source.
Samples of straight oils, oil lumps and tarballs, oily water, oil emul-
sions, and oiled surfaces (rocks, shells, coarse sand, wood, cleanup straw,
brooms, etc.) are easily processed. The soluble portion of these samples
is often contaminant oil, with little or no admixed natural hydrocarbon
background. These samples are suitable for routine analysis and attempted
correlation. These materials can be monitored, often at ppm levels, for the
presence of oil or oil products.
Concerted efforts to devise and standardize procedures for extracting
and detecting low levels of oil in organisms and in sediments have been made.
Appendix E is a preliminary summary of these efforts, made under the spon-
sorship of the International Decade for Ocean Exploration (IDOE) at the Woods
Hole Oceanographic Institution and elsewhere. More detailed reports have
been published and can be consulted in this regard. The SCOT GC columns em-
ployed in this study are capable of analyzing the materials obtained by these
sample-preparation procedures.
OIL-RICH SAMPLES
Oil samples or samples which are predominantly oil (such as "tarballs,"
very oily sand, and thick deposits removed from surfaces such as rocks,
shells, pilings) are prepared by dissolution in carbon disulfide and sepa-
rated from water and particles by centrifugation. All carbon disulfide and
other solvents used are reagent grade, redistilled in all-glass apparatus
(CAUTION: CS is highly toxic and has a flash point of -22°C) , and shown by
100X evaporative concentration and GC analysis of the residue to be free of
interfering materials.
18
-------
The thawed oil rich sample is shaken and stirred or otherwise homogen-
ized, and a portion (containing approximately 50 to 100 mg of oil) is placed
in the bottom of a centrifuge tube equipped to accept Teflon-lined screw
caps. Thin oils can be transferred by dropper or capillary; thicker ones,
on a rod. Sufficient carbon disulfide is added to result in an estimated 5%
to 10% solution (for pure oils, it is convenient to weigh oil and measure CS_
volumetrically so that the concentration is accurately known). Heavier and
darker oils generally require 10% solutions, but this is not critical. The
use of 50 to 100 mg of oil and 1.0 ml carbon disulfide in a 13-ml centrifuge
tube that can be closed with a Teflon-lined screw cap is recommended. The
sample is agitated for several minutes, as with a Vibromixer." All the com-
ponents of interest in oils are soluble in CS~, and all but the thickest oils
dissolve rapidly; thick oils should be smeared as a thin film on the centri-
fuge tube walls.
Very thick oil samples may be warmed in Teflon-capped centrifuge tubes
to about 50 C and agitated. The sample is then centrifuged to remove insol-
uble particles and water globules. Removing about half of the liquified oil
from mid-depth with a syringe equipped with a long needle leaves behind
supernatant bulk water and solid residues. The oil sample is then dissolved
in CS~ as previously described.
The resultant CS~-oil solutions are stable for months if chilled and
protected from light and evaporation by storing in glass vials with Teflon-
lined caps in the dark. If possible, store samples in this manner rather
than as frozen oils, which may precipitate waxy deposits difficult to redis-
solve.
OILY WATER, SAND, STRAW, ETC.
Samples can be treated by the procedure given above after a simple step,
such as scraping the oil coating off shells. A common sample is water of 1%
to 10% oil content with some solid material present. Centrifuging such sam-
ples for 20 minutes at 2,400 g (clinical type centrifuge) often yields well-
defined surface or sunken layers that allow the removal of water. The oil
residue is treated as an oil-rich sample; if oil collects at both the top
and the bottom, both portions should be recovered by pipetting or syringing
out the water. More dilute oily water (1:1,00 or more) is extracted with
several small volumes of carbon disulfide in a separatory funnel.
If solutions are too dilute (estimated by trial GC analysis or by evap-
orating an aliquot and weighing the involatile residue), they may be concen-
trated on a rotary evaporator by controlling the suction at above 100 mm Hg.
Full mechanical pump vacuum is applied for 30 seconds at the end of the
evaporation to yield nearly solvent-free material. This method removes most
of the material boiling below n-C^ but hardly affects components boiling
above n-C, £..
19
-------
SEDIMENTS AND ORGANISMS
Sample Extraction
These samples require more elaborate cleanup procedures and parallel
determinations of the indigenous background of pre-spill samples. Select
Soxhlet apparatus, extraction flasks, separatory funnels, and chromatography
columns compatible with the relative volumes specified. All apparatus and
solvents must be purified before use. Pentane, benzene, and methanol are
distilled in all-glass equipment through a packed column and checked for res-
idue (see CS. procedure). Granular anhydrous sodium sulfate, boiling stones,
paper Soxhlet thimbles, and filter paper such as Whatman No. 1 are purified
by 24-hour Soxhlet extraction with 1:1 benzene-methanol, air dried in a
clean hood, and oven-dried at 110 C. The adsorbents silica gel (equivalent
to Davison chemical Grade 922, 200 to 325 mesh) and alumina (equivalent to
Harshaw Chemical Company, AL0102P) are Soxhlet extracted with 1:1 benzene-
methanol, air-dried, activated (silica gel, 120 C; alumina, 250 C) and then
deactivated with water (silica gel, 5% w/w; alumina, 6% w/w) by thorough
mixing and 24-hour standing. Prepared adsorbents may be stored in tightly
stoppered, all-glass bottles. Adsorbent activities and consequently chro-
matographic results vary from batch to batch of adsorbent. Chromatographic
procedures are not entirely standard and should be checked for each new sam-
ple type and reagent batch.
The thawed sediment or organism sample (50 to 100 g is convenient) is
loaded into a Soxhlet thimble (not more than three-fourths full). If sample
crushing or cutting is necessary to reduce the size, clean, solvent-insol-
uble surfaces and apparatus must be used. For organisms, specify the ma-
terial taken (e.g., "whole organism", "shucked shellfish" (with or without
internal fluid), "muscle only").
The thimble is set in the Soxhlet extractor body on a large glass
stopper or plug to facilitate good drainage, and extracted with 1:1 benzene-
methanol (about 4 to 5 times extractor body volume). The solvent level in
the boiling flask should never fall so low that the dry flask wall is heated
directly. The extracting solvent is changed after several hours. If heavy
foaming occurs (as with some biological materials), an earlier change or two
changes may be required. The total extraction time should be at least 24
hours. Stopping extraction just before the Soxhlet cup siphons minimizes
volumes to be handled.
The extracted material is then prepared as a solvent-free lipid frac-
tion and weighed before column chromatography. This is accomplished by the
following procedures.
A. If the benzene-methanol organic extract contains only small amounts of
particulate matter and if a small test portion of extract does not emul-
sify on diluting 5:1 with distilled water or with half-saturated sodium
chloride solution (previously pentane extracted to remove contaminants),
the procedure is straight forward. The entire benzene-methanol extract
is then decanted from solids or filtered through paper, diluted with
20
-------
water, extracted 5 times with pentane (in total, at least one-fourth of
the aqueous volume). The combined pentane layers are dried over sodium
sulfate and filtered. The sodium sulfate and filter were previously
extracted with pentane to remove contaminants. The flask and drying
agent are rinsed with fresh pentane. The pentane of the combined ex-
tracts is then evaporated and the residue of the extracted material is
weighed.
B. If the benzene-methanol extract contains much particulate matter, the
extract is centrifuged. The solid portions are re-suspended in pentane,
agitated, and recentrifuged. The pentane layer is drawn off and added
to the pentane used to extract the diluted benzene-methanol supernatant
layers as described above in Part A. It may be convenient to extract a
known large fraction of the extract to keep required volumes compatible
with available glassware. The hydrocarbon content of the total ex-
tract is then calculated by proportion from the extracted subsample.
C. If the water dilution test shows the formation of emulsions that do not
break in about 30 minutes (some biologically derived samples), diffi-
culties may be encountered. The trial portion is tested to see if high-
speed centrifugation will break the emulsion. If the emulsion breaks,
the remaining benzene-methanol extract may be diluted with water and
centrifuged. Otherwise, the benzene-methanol extracts may be repeatedly
(five times) treated with relatively small volumes of water or salt so-
lution and stirred slowly but thoroughly to minimize emulsification.
All five water extracts are saved and back-extracted with pentane. A-
cidification of extracts to pH 2 may minimize emulsification, especially
for samples rich in free fatty acids. Organism extracts may also be
treated by saponification with methanolic KOH, subsequent dilution with
water, and extraction of the nonsaponificable materials (including pe-
troleum hydrocarbons) into pentane. See Appendix E and references
therein for further information. Once the initial phase separation and
extraction of material is accomplished by any of the above methods, the
sample is dried, concentrated, and weighed.
Column Chromatography
All sediment and organism-derived extracts must be purified by column
chromatography before gas chromatography. Sediment extracts also require
removal of elemental sulfur by reaction with powdered copper. The dried and
concentrated sample extracts (the entire extract or 0.5 g, whichever is
smaller) are redissolved in a minimum volume of pentane and placed on the
chromatography column. The samples are then chromatographed by eluting the
column with four column volumes of n-pentane. The columns are prepared by
wet-packing 60 times the sample weight of silica gel in pentane and covering
this with an additional 40 times the sample weight of alumina. Both adsor-
bents are purified and deactivated as previously described in the preceding
section. The chromatographic tube should be chosen such that the packed bed
height is 15 to 25 times the internal diameter. The available columns often
determine an optimum adsorbent weight and the maximum acceptable sample
weight. These water-deactivated dual-absorbent columns have a high capacity
21
-------
for high-molecular weight non-hydrocarbons, and they do not tend to convert
sensitive nonhydrocarbons to artifact hydrocarbons. These columns have been
found to give 99 + 1% recovery of fuel oil hydrocarbons ( a carbon number
range that covers the most commonly utilized peaks for oil correlation).
The suitability of the column for removal of nonhydrocarbons is checked
for each batch of adsorbents and each new sample type. The check is per-
formed by chromatographing a representative oil of the sample or a portion
of the original oil sample. A thick film infrared spectrum on NaCl plates
of the concentrated column effluent (micro-apparatus may be required) should
show no -OH, C=0, or cis- and trans- di-substituted double bond absorption.
Presence of these bands establishes that polar materials have broken through
and requires that column parameters be checked and samples be rechromato-
graphed.
The solvent is stripped from the chromatography column effluent and the
concentrated column residue is weighed. The material in the Soxhlet cup is
air-dried for several days with occasional turning and also weighed. Adsor-
bents are discarded after a single use. Following concentration of the col-
umn effluent (i.e., solvent stripping), CS^ solutions of the residue are
analyzed by GC. The hydrocarbon content may be calculated as the chromato-
graphy column effluent weight divided by the dry weight of the extracted
residue. However, the residue is not necessarily all detected by GC anal-
ysis, and the gas chromatogram cannot be assumed to represent all of the
injected material.
Removal of Sulfur from Sediment Samples
Sediment-derived samples require removal of elemental and reactive sul-
fur, which passes through the alumina-silica gel column before GC analysis.
The sulfur content of sediments is often quite high, and this step cannot be
neglected. Electrolytic copper dust (25 times the weight of the extract to
be treated) is covered with concentrated hydrochloric acid and stirred on a
sintered-glass funnel. The acid is filtered off without drawing large
amounts of air through the copper. This washing may also be performed in a
centrifuge tube. The cleaned, bright copper is washed with successive
portions of water (two times), methanol (three times), benzene (one time),
and pentane (one time), draining well after each washing.
The Cu powder is transferred in pentane to a chromatographic column.
The packed Cu bed should be at least 20 times as long as the column internal
diameter. The column is washed with several bed volumes of pentane, and the
sample is slowly percolated through the column and rinsed with 4 bed volumes
of pentane. The removal of sulfur is accompanied by blackening of the upper
part of the column. If more than the upper third of the column blackens,
the sample should be treated again. If no blackening occurs, the column
should be tested with a small amount of pentane solution of elemental sulfur.
If no blackening then occurs, the column was inactive and the sample must be
retreated. These columns should be freshly prepared. They may be used for
several samples on the same day if thoroughly rinsed.
22
-------
PREPARATION AND USE OF STANDARD REFERENCE SAMPLES
Analysis of standard samples is an essential part of the method. The
procedures for analyzing oil are complex, possibilities for contamination
are numerous, and the variations in instrument and column performance are
significant. Some standard of relatively fixed and reliable properties is,
therefore, of great value. A standard may be utilized to check the perfor-
mance of analyst, instruments, and reagents, and is invaluable in locating
the source of difficulties. Chromatography columns wear out, and a standard
is required if results obtained from more than one column are compared.
Standard No. 2 Fuel Oil
A standard No. 2 fuel oil can be prepared by column Chromatography of a
commercial No. 2 fuel oil on silica gel/alumina without solvent to remove the
light-sensitive and sediment-forming, dark-colored material. GC traces of
our standard No. 2 fuel oil are presented in Appendix D. The chromatogram
can be recognized by the presence of an intense peak eluting just after the
CS peak. The standard oil sample may be used routinely to check the perfor-
mance of instruments and columns. Laboratories requiring larger amounts
(hundreds of injections) may easily prepare a suitable secondary standard.
Standard Sediment Sample
A standard sediment sample may be prepared by obtaining sediments from
a relatively nonpolluted area. A sample of sand-mud and mud sediments would
be appropriate. Care should be taken to avoid contamination of the sediments
by exogenous oil contaminants from samples, etc. The sample is then spiked
with a known quantity of the standard No. 2 fuel oil. The sediments should
be homogenized with the standard oil in a solvent-cleaned, oil-free homogen-
izer for approximately 15 minutes. Aliquots of the sample can then be trans-
ferred to glass pint jars with tin-lined caps and frozen until analysis.
A standard sediment and organism sample can be prepared for standard-
izing and comparing all phases of the sample extraction and preparation pro-
cedures, including the gas Chromatography and peak-height measurement. Ap-
pendices C and E describe a sampling procedure and results in which sediment
samples were utilized from an area polluted by the West Falmouth Oil Spill
(Blumer elt al. 1970, 1972).
Standard Organism Sample
A simulated standard organism sample can be prepared by mixing a 0.2%
(w/w) solution of the standard No. 2 fuel oil in commercial cod liver oil.
This should be stored under nitrogen in the cold. An aliquot of unspiked
cod liver oil should also be frozen. The cod liver oil used in this study
contained approximately 500 micrograms/g of hydrocarbons passing through our
workup procedure. Squalene and pristane were the major components in the
natural hydrocarbons. The added oil was 2,000 Pg/g. Appendix E reports the
results of an intercalibration exercise using these standards.
23
-------
ARTIFICIAL WEATHERING
Evaporative weathering can be simulated in the laboratory. The fol-
lowing procedure simulates evaporative losses of source oils (e.g. Appendix
D, chromatograms 7, 8 and 9).
The oil sample to be weathered is placed in a flask and connected to a
rotary evaporator. Vacuum is applied slowly until a full mechanical pump
vacuum can be applied without excessive foaming or bubbling, and evacuation
is continued while the temperature of a water bath surrounding the flask is
raised slowly. Small samples are withdrawn periodically for GC analysis.
Several hours of pumping at below 1 mm Hg in a bath at 60 C or warmer may be
required to yield much evaporation in the GI? region.
An oil sample that resembles the environmental oil sample in degree of
evaporation is utilized. This similarity is judged by the comparison of
shape of the low boiling range signal or a ratio of background at C, ,. to any
signal parameter at C^7 or C,~. The ratio differences are tabulated to de-
termine the shift in analytical ratios caused by evaporative weathering.
This method should be applied with caution if ratio changes greater than 20%
are required. These samples are heavily weathered. A newer method was doc-
umented in 1975 as an American Society for Testing and Materials (ASTM, 1975)
procedure. This is method No. D3326-74T.
24
-------
SECTION 7
SAMPLE ANALYSIS
PREPARATION OF GAS CHROMATOGRAPH AND COLUMN
Varian model 1440 gas chromatographs equipped with capillary injector
kits were utilized exclusively in the development of this method. Other gas
chromatographs can be utilized, but necessary modifications or problems they
may present cannot be specified. Capillary injection design varies widely.
Set-up and checkout procedures specified in the instrument manual
should be performed. In addition, the indicating pyrometers and temperature
programmer settings of various gas chromatographs require checking. An
accurate (+ 1 C) calibrated thermocouple or other temperature element should
be used to measure the temperatures in the injection port, the oven at
various temperatures, and the injector base. A thermocouple in a clean
glass capillary tube sealed at one end should be used. The instrument pyro-
meter should be calibrated and the isothermal and temperature-programmed
limits checked and adjusted (see Instrument Manual). The accuracy of the
program rate at 6 C/min (similar rates of 5 or 7.5 C/min are acceptable) is
checked by timing and observing the dial rotation. Uneven rotation of the
mechanical slip-clutch, which connects the step motor to the program rate
potentiometer, has occurred in several of these instruments.
An easily replaced borosilicate glass liner is installed in the cap-
illary injector. Figure 1 illustrates the shape and position of the liner
in the injector. The tubing is 55 mm long, 2 mm "Pyrex" tubing. The end
flare illustrated must be large enough to guide the needle through the tube,
but not so large as to form a seal against the injection port's flare or the
septum. The central slotted tube of the injector may be irregular in form
or have a burred edge. The tubing can be rounded and deburred by gentle
-n.-'nipulation with a small circular file. There is no danger of injuring the
split orifices by this procedure. It is not accessible from the injection
port. Many current GC instruments are equipped with adequate glass liners.
The gas chromatograph should be connected according to manufacturer's
specifications to a 1-mv full-scale recorder of 17.5 to 25 centimeter
(7 to 10 inch) chart width with better than 1 sec response time and small
deadband. Leeds and Northrup Model H recorders were used. If chromato-
grams are to be photographed as evidence, choose an ink which photographs
well (dark red). The uniform use of a chart speed of 1.27 centimeter (one-
half inch) per minute is strongly recommended, with signal increasing with
pen movement to the right. The resulting chromatograms read from right to
25
-------
left, "backwards" from the usual convention. They are not easily confused
with traces run coventionally. Comparisons of tabulated data do not depend
on recorder speed or format, but detailed visual comparison is much easier
if traces are all of the same signal polarity, temperature program, and
chart speed.
The columns are SCOT columns from the Perkin-Elmer Corporation, Column
Coating Facility, Norwalk, Connecticut. They may be obtained by ordering
"50 foot x 0.02 in. SCOT columns coated with OV-101, to fit Model (instru-
ment used) GC; coated without use of wetting agents, and rating 25,000
effective plates on the Perkin-Elmer hydrocarbon standards." Two slotted
male SS 1/16 inch- nuts, Perkin-Elmer #009-1593, are required to connect each
column to a Varian 1440 with a capillary injector. The three columns des-
cribed in this study were supplied within 6 weeks of ordering them. The two
columns ordered simultaneously were most similar. Though a single column
has a fairly long lifetime (apparently at least 200 oil --'njections) , we
recommend ordering them in pairs. The feasibility of some central group
obtaining larger batches and conditioning and intercomparing them should be
considered.
The columns were preconditioned at 160 C by the manufacturer and are
further conditioned with the column disconnected from the detector. The con-
ditioning schedule in Table 4 yields columns that still change bleed char-
acteristics fairly rapidly over the first 30 or so injections. The initial
period may be used to familiarize the operator with the method. Tempera-
ture-programmed column stationary phases became conditioned differently
during repeated programming than isothermally. We recommend periods of iso-
thermal operation at 275 C interrupted with repeated programming sequences
for any further conditoning desired. This conditioning procedure and that
in Table 4 are illustrative and do not need to be followed exactly.
Table 5 summarizes the operating parameters utilized for analysis; in-
strument settings should be adjusted accordingly. The methane retention
time (MRT) is measured by stopwatch as the time lag between injection of
about 1 yl of methane (burner gas stored in a septum-covered flask) and
the initial recorder pen rise. This provides an accurate measurement of
carrier gas flow rate without disconnecting the column or using inaccurate
flowmeters often supplied with gas chromatographs. If a rapid, smooth in-
jection is not made, methane will exit in multiple peaks corresponding to a
series of closely spaced mini-injections. If this occurs, the procedure
should be repeated.
Injection splitters were employed in our instrumentation, but good
resolution may be obtained with SCOT OV-101 columns in modern GC instruments
without injection splitters if the detector dead volume is small and is
swept with make-up gas. Samples (0.05 to 0.5 yl) may be introduced by a
1-yl syringe (these require careful cleaning) or as films (10 to 20 yg) on
a glass rod (Ehrhardt and Blumer, 1972). The latter technique distorts the
chromatogram in the noncritical (below n-C^-) range. Although resolution
is adequate with non-splitting conditions, we have not determined the short-
er long-term reproducibility attainable. The RSD's of ratios of widely
26
-------
separated components may possibly even decrease without the splitter.
TABLE 4. CONDITIONING SEQUENCE OF SCOT OV-101 COLUMNS'
Duration
Temperature, C
Carrier gas (He) flow
2 hours
2 hours
Several standard
No. 2 Fuel Oil injections
16 hours
30 minutes
Five to ten injections
of #2 FO
1 hour minimum
220U
220°
Program 75-250C
@ 6 /min
225°
250
Program to 250
275^
6 ml/min
6 ml/min
Adjusted for CH
RT = 44 sec at 75°C
Adjusted for CH,
RT = 44 sec at 75°C
Several temperature
programs
75° - 275°
@ 6 /min
Adjusted for CH,
RT = 44 sec at 75°
a Column not connected to detector.
b See daily operating procedures.
c Column usable but further operation at 275 C increases baseline
stability.
The instrument setup is completed by running a series of standard No. 2
fuel oil analyses until the analytical ratios become essentially constant,
and then estimating ratio RSD's of five to eight runs. Table 2 contains
the results for several columns. Similar RSD values should be obtained.
When the same standard No. 2 fuel oil is used, similar absolute ratios and
a chromatogram similar in appearance to those in Appendix D should be ob-
tained. Exact coincidence of fingerprints are not expected.
27
-------
TABLE 5. GC OPERATING CONDITIONS FOR OIL ANALYSIS
Item Conditions
Injector 195 to 210 C;use silicone septum and glass liner.
Helium flow 50 ml/min, measured at split exit.
Column Temperature programmed from 75 to 275 at G°/min,
and with isothermal operation at 275 up to 35 min.
Oven He carrier gas; flow rate adjusted to a methane
retention time (MRT) of 44+0.5 sec at 75 C.
Detector Flame ionization operated at 300 + 10°C with
10 ml/min He makeup gas and optimized compressed
air (not pumped lab air) and hydrogen flows.
Electrometer Sensitivity 1 x 10 amps/mv.
Injection 0.5 to 5.0 yl of CS_ solution by syringe.
Recorder 1 mv, 17.5 centimeter (7 inch) or greater span, 1-second
response; operated at 1.3 centimeter (0.5 inch)/minute
chart speed with response positive left-to-right.
ANALYSES AND DAILY OPERATION
At the beginning of each day and/or series of analyses, the following
procedure is followed. The glass tubing liner is changed by unscrewing
the septum nut, pulling out the old liner, and inserting a clean untouched
one. A return from standby to active condition is achieved by turning on
the air and hydrogen flows, increasing the He pressure from a few Kg/cm to
457 Kg/cm (65 psi) (or recommended pressure for instrument used), igniting
the flame, turning on the recorder, turning on the flame voltage, and
setting the sensitivity switch to 1 x 10 amps/mv from "balance." The col-
umn is heated from 75°C to 275°C in 15 to 20 min. and cooled to 75 C; the
methane retention time (MRT) is then measured. If necessary (rarely) the
carrier gas flow is adjusted to maintain the MRT at 44 + 0.5 seconds by ad-
justing the split ratio.
The instrument "balance" is set and checked according to the manufac-
turer's recommendations. Pen deflection should be very small (50 micro-
volts on the 1 mv scale). Only if the pen returns accurately to the same
signal at 75 C without variation can baselines be applied to sample chro-
matograms. Failure of accurate return may be caused by a dirty detector,
excessive column bleed, or too short a cooldown (5 to 10 minutes minimum).
The small reproducible signal at 75 C is damped.
28
-------
Samples, baselines, and standards should be analyzed in a sequence that
facilitates making comparisons by placing similar samples together. This is
accomplished by analyzing similar samples separately, evenly spacing base-
line determination of standard No. 2 fuel oil runs (or any other standards).
For very close comparison, samples are analyzed with sample size adjusted to
yield comparably sized chromatograms, determined by a sample-size/peak ratio
curve as in Figure 2. A maximum signal intensity of 0.5 to 1.0 mv is sat-
isfactory for routine work. If the sample size recommendations regularly
yield too large or small injections, the split gas flow should be increased.
The MRT must be kept at 44 seconds.
Each analysis is initiated by starting the recorder chart drive and in-
jecting a solution with a 5- or 10-yl syringe as the pen approaches a chart
division. Mark the chart by momentarily moving the electrometer range
switch division, while the needle is still in the injector. Injection tech-
niques vary; be consistent. Withdraw the syringe plunger by half the needle
volume after loading the syringe so that an air plug fills the front half of
the needle. Sample fractionation occurs if the full needle is brought near
the hot injection port. Avoid injecting excessive volumes of air.
Start the program when the solvent peak begins to elute. After the
program is completed, maintain 275 C until the detector signal returns to
the level of the previously determined baselineor for 35 minutes whichever
is sooner. Cool the column back to 75 C, reset the programmer, cut the
chromatogram off, and hang the weight (if one is used) on the recorder chart.
The recommended standby condition is: air and hydrogen off, helium pressure
reduced to a few pound (CAUTION: gauge readings stabilize slowly. Do not
turn off completely), flame voltage off, and electrometer at "balance." The
recorder electronics may be left on or off. The oven is set at 75 C. For
extended periods, disconnect column from detector.
MAINTENANCE PROCEDURES AND STANDARDS
The following procedures are performed when needed. Replace the septum
every 23 injections or when needed because of leaks, erratic MRT, or bleed
peaks. Precondition septa to minimize bleed by storing them in the column
oven in a beaker.
Clean the detector as described if symptoms of dirty detector appear,
especially "spiking" or failure to return to low signal level after cooling
the column back to 75 C.
Replace the injector liner and clean the dirty one in CS~ every three
or four injections, or more frequently if high-residue oils (crudes, as-
phalts, heavy fuels) are being analyzed.
Clean the injector by the procedure given in Appendix F at intervals of
100 injections, or whenever RSD's of ratios involving peaks of different
boiling points increase. The increased variability is a sign that tars or
glass fragments are in the injector tubing and require flushing out. If the
instrument is idle longer than a weekend, turn injector heaters off to mini-
mize bleed onto the column. For extended periods, disconnect column from
29
-------
detector.
As the column bleed stabilizes, an occasional (every 60 or so injec-
tions) series of replicate standard No. 2 oils should be run four to six
times to establish short-term RSD values and to compare with earlier perfor-
mance. The standard oil should be run at least every 20 injections.
Determine a baseline curve (CS? injection, then program) every 25 in-
jections or when isothermal signal at 275 C changes 5% or more.
CHROMATOGRAM MEASUREMENT AND DATA TABULATION
Comparison of details of chromatograms requires direct visual compari-
son of traces obtained under comparable conditions. However, measuring and
tabulation certain signal intensity ratios from gas chromatograms abstracts
most of the useful information in a form suitable for simultaneous compari-
son of many samples and for statistical purposes. The procedure is de-
scribed below. All useful chromatograms should be tabulated.
A baseline curve is first superimposed on the sample chromatogram of
interest. The vertical distances at selected points between baseline and
chromatogram signal are then measured and recorded, and finally, signal in-
tensity ratios are calculated and tabulated (Figure 3 and Table 3). The
baseline curve should be obtained from the same instrument/column combina-
tion. When the column and septa are well conditioned, and if the daily
bake-out procedure is followed, the baseline should be a smooth curve devoid
of peaks. The absolute signal level ( in mv) at the end of the temperature
program should be constant. Within a few percent, the same signal level
should be asymptotically approached at the end of a sample chromatogram as
the sample elutes isothermally. These criteria are generally satisfied for
a period of 25+ injections by a single curve. A new baseline should be de-
termined after each septum change to guard against effects from materials
vaporizing out of the septum. Ratios should not be determined at locations
at which septum bleed peaks appear without extremely frequent baseline
checks (cold columns collect septum bleed). Septum bleed peaks have occurred
at considerably higher elution temperatures than the signals we ratio in our
work.
The baseline curve is traced accurately onto the sample chromatogram
using the rising edge of the solvent peak as a fiducial mark and further
aligning the chromatograms by the 0% and 100% scale lines. It is essential
that programs be started uniformly, that programmers track correctly, and
that recorders feed paper without slippage. We have found that mechanical
slippage within programmers may occur. The programmer should be timed fre-
quently during at least one full revolution of the indicating dial to see
that it advances at a reproducible rate.
Accurate vertical distances along the millivolt axis are measured with
a glass and ruler calibrated in hundredths of an inch, with care to avoid
parallax error. Different observers rarely disagree by more than + 0.25 mm
(0.01 inch). The most useful ratios require measuring the heights of the
30
-------
e
o
03
O
en
c
i
o
CO
E
CD
05
2
co
o
O
co
45
-1H9GH MV3d 3NVlAHd
31
-------
n~C17' Pristane> n~ci«' and Phytane Peak above the traced baseline. These
peaks elute in the order mentioned in our system and can often be recognized
visually in a chromatogram (Figure 1 and Appendix D). They may also be lo-
cated to within + 1 C by measuring the distance to the solvent peak and com-
paring with the peaks of the prepared standard fuel oil or another reference
sample such as Bachaquero crude oil. These peaks are inconspicuous in such
rare oils as Bachaquero crude and the sample may then be re-analyzed with a
small added spike of the prepared standard (Appendix D, Chromatogram 15).
The analyst should recall that the designations of these peaks are
names and not strict chemical identifications, and also that signal inten-
sity measurements, rather than area measurements, are made. It is erroneous
to view the ratios as being accurately related to the amounts of the named
substances present in the sample. Though it is convenient to form four of
the possible ratios, only any three of the four are independent parameters.
A measure of the extent to which the signal consists of a very complex
"background" of unresolved substances may be tabulated. A convenient but
arbitrary measure is formed by drawing a tangent line to the valleys immed-
iately preceding the n-C,7 and n-C1fi peaks, and measuring the distance from
the baseline to this line along the vertical that passes through the n-C17
peak. We recommend replacing this ratio, or supplementing it, with a ratio
based on the vertical that passes through the phytane peak, the tangent line
being extended slightly to underly this peak. This latter ratio is less
subject to alteration by evaporation weathering or by bacterial action on
theoil than is the n-C17 background ratio. The ratios should be formed to
four significant figures, but rounded to three unless they are to be aver-
aged or utilized in statistical estimates. Further ratios may be found
useful in particular applications.
When two chromatograms do not differ with respect to a feature by visual
inspection, they will rarely differ significantly when measured and tabu-
lated. Samples often differ very widely in their overall boiling point dis-
tribution even through they are similar in some region. A sketch of the
overall envelope shape in the "remarks" column of a tabulation sheet will
often suffice instead of creating new ratios to measure this.
Within the sample sizes recommended, the sample size has only a small
effect on ratios. Figure 2 summarizes the results of a study of sample size
vs. ratio values performed on column 219 using the prepared standard No. 2
fuel oil. For work with very similar samples or where multiple runs are u-
tilized, sample size-ratio regression lines could be utilized to increase
accuracy, particularly for the peak/background type ratios that are most
sensitive. More information could be gained by standardizing sample prepa-
ration and introduction so that absolute signal sizes could be accurately
compared. However, this would require considerable effort with little pro-
bable gain.
When large numbers of samples are analyzed with little foreknowledge of
their type, many traces may have poor signal intensities. An estimate of
sample sizes can be achieved on a large number of samples by running an old
32
-------
column isothermally at high temperature and injecting the samples (injector
50 C hotter than oven) at fairly short time intervals in another instrument.
An estimate of the proper injection sample size vs. signal intensity reso-
lution can then be obtained, although the column used will deteriorate rap-
idly and the instrument injector/splitter will require cleaning.
33
-------
SECTION 8
INTERPRETATION OF DATA
This section presents detailed guidelines for correlating weathered
oils with suspected source oils by gas chromatography.
Situations frequently arise in which no single clear correlation exists
source oils do not match the spill, or several indistinguishable possible
source oils exist. Situations leading to false correlations occur when an
environmental sample that contains a mixture of oils or oil control chem-
icals is mistaken for a single oil and the ratios are correlated with a
possible source. Unusually intense bacterial alteration can affect ratios,
especially those involving n-hydrocarbons. Occasionally, the actual source
is missing but the sample is correlated with a very similar suspected source.
The following guidelines are designed to minimize such false correlations.
The major factors the analyst must always consider are the completeness
and validity of the sample collections, the effects of weathering on the en-
vironmental sample, and the ability of the analytical technique to distin-
guish different oils. The following list shows several types of problems
that may be encountered.
Unrepresentative samples
Missing sources
Unrepresentative source samples
Unrepresentative spill samples
Admixed spill control chemicals
Mixed spills
Mixed with indigenous hydrocarbons
Unresolvable pairs
Multiple samples of the same oil
Similar oils, especially crudes
Inadequate resolution
Environmental alteration
Unusual bacterial alteration
High levels of indigenous hydrocarbons
Mixtures
Severe evaporation or old samples
Unknown processes
34
-------
The distinguishing power of the technique can be estimated and is
therefore the least serious problem. Estimates of the RSD's of various
ratios can be made by replicate analyses. The RSD values of ratio-means
for the standard oil apply approximately to the sample. By analyzing the
pr Unary standard samples or secondary work standards, the analyst checks
the instrumental and measurement steps and most major sample preparation
variables. Information gained by these tests can be utilized either qual-
itatively by subjective evaluation or quantitatively by applying simple
statistical tests to judge whether observed differences in sample ratios
are significant or have occurred by chance. We have utilized Student's t
test to compare data.
Sample collection validity and weathering effects are not subject to
such rigorous control. We recommend that routine correlation of oils with
sources be limited to samples that are young and oil-rich enough so that
the listed problems are minimized. Oils, oil slicks, "tarballs," oil-rich
water, and oil collected as discrete coatings on sand, sediments, and or-
gansims, etc. will generally have a negligible indigenous substrate hydro-
carbon content. These larger masses of oil are also unlikely to be strongly
affected by bacteria or to become thoroughly mixed with other oils in short
time periods. Oils estimated to be less than a month old are most promising
for correlation. The source collection for older oils is likely to be com-
plete, and weathering effects will be more pronounced. A large percentage
of the situations of interest in water-quality management are weeks-old,
oil-rich samples suitable for routine work.
In contrast, oil extracted from dilute water samples, sediments, and
organisms is more likely to be significantly weathered and accompanied by
an admixture of indigenous hydrocarbons. These samples require preliminary
separations and parallel background determinations on control samples. Ex-
tensive and intensive sampling programs may be required to estimate the ex-
tent of weathering. The probable source collection is difficult to com-
plete if much time is elapsed. We recommend that only laboratories special-
iz_injg_ in oil analysis consider such samples for routine correlation. The
analyst employed in this work must be capable of applying the general cri-
Ler ia set forth and have a sound knowledge of oil chemistry and the chem-
istry of environmental lipids.
However, it is possible to establish routinely, the presence of oil in
sediments and organisms even when correlation with specific sources of
chemical evidence alone is unwarranted. This presence may be important
after a major spill in establishing the fate and effects of oil not re-
quiring precise correlation.
For routine analysis of samples, a set of chromatograms of the envir-
onmental sample and all available possible sources is required. This data
set should be of the maximum feasible camparability, especially if similar-
appearing samples are involved. For close comparisons, a standard sample
of No. 2 fuel oil and a second subsample of the spill (not re-analysis of
the same solution) should be run so that the entire group is bracketed by
these analyses. If alcohol-type spill control chemicals have been used, the
35
-------
samples must all be column chromatographed as described for organisms
and sediments. Hydrocarbon-base spill chemicals cannot be removed. In fa-
vorable cases, mixed samples may be treated as a linear combination to de-
rive ratios for the spilled oil.
If the two chromatographic fingerprints of the subsamples of spilled
oil are similar within the expected analytical error, correlation can be
demonstrated. Otherwise, either the performance of the method or the homo-
geneity of the sample are deficient. Check the method performance by anal-
yzing the prepared standard No. 2 fuel oil and remedy any defects. Analysis
of additional subsamples or sample recollection may be necessary.
The mean values of the analytical ratios for the two spill sample gas
chromatograms are calculated, and the onset of appreciable signal in the
spill sample chromatograms is used to judge the volatility losses. If the
onset of signal in the environmental sample has actually been determined by
the composition of the original oil, an exact correspondence of spill and
source chromatograms will exist. The source oil will then be material free
of low-boiling components. If the onset of spill sample signal is below
n-C..,, none of the tabulated ratios should be affected by evaporation. On-
set of signal at or below n-C implies alteration of the ratios which can
be simulated by artificial weathering. Onset of signal beyond n-C,, implies
relatively strong alteration of ratios other than C /pris, C _/phy, and
ratios to background signals at higher boiling points. Such evaporated
samples are probably more than several weeks old.
The environmental samples is compared with each suspected source in
turn. The spill and its source should match after the allowances suggested
above are taken into account. It is particularly important to compare all
sources and not to identify the spill sample with the first matching source
encountered until all others are ruled out. The "possible source" list is
narrowed by rejecting samples with any markedly different (5 to 6 times the
RSD) peak-height ratios, different boiling-range distributions (envelope
shape), or patterns of smaller peaks. These refections narrow the field,
and in simple cases it narrows the possible sources to one. Generally, the
rejections drastically narrow the possible source oils. If further compar-
isons are necessary, repeat analyses of the more similar oils and determine
whether significant differences between source and sample exist according
to some statistical measure such as Student's t test. At this stage, it
may be desirable to artificially weather by evaporation (Section 6) the
candidate sources for closer comparison of the C.. /C10 and pristane/phytane
LI -Lo
ratios.
We define a unique correlation as one for which all but one possible
source is ruled out, and one source does not differ significantly (after
artificial weathering if necessary) from the environmental sample. Repli-
cate analyses for better comparison may be performed.
For samples that seem to correlate except for ratios involving the
n-hydrocarbons, the possibility of unusually rapid microbial catabolism
must be considered. This will lower the n-hydrocarbon peaks by a nearly
36
-------
uniform amount over the entire chromatogram and thus lower the correspond-
ing ratios. Microbiological attack on other oil components is much slower,
and correlations established with at least one nonnormal ratio are corres-
pondingly more secure.
Each correlation should be reviewed on the basis of the assembled evi-
dence. The possibility that any of the possible sources was unusually ra-
pidly attacked by microorganisms, lowering the corresponding ratios, should
be considered. The other features (envelope shape, small peak patterns) of
the chromatograms should be checked carefully to confirm that no information
contradicts the assignment made.
We have inserted cautionary comments repeatedly about n-hydrocarbon
oxidation by bacteria. This is primarily because laboratory experiments
have indicated that rapid (days) alteration of oils is possible. On the
other hand, the weight of evidence suggests that the ideal conditions re-
quired are rarely achieved in the natural environment. Laboratory experi-
ments often involve conditions rare in nature: high nutrient concentra-
tions, mixed cultures, aeration, the availability of additional energy
sources, and strong agitation to disperse the oil. Oil films on hay or de-
caying seaweed may naturally alter relatively rapidly. However, the alter-
ation of larger masses (centimeter dimensions) of oil in a few weeks time
has yet to be observed.
Finally, the analyst should check his conclusion against adjunct in-
formation about the spill, its possible sources, and the conditions which
the sample withstood. With experience, he can streamline procedures to min-
imize duplicate analyses and compare samples with known locally abundant ma-
terials. The ratio stability and comparability afforded by the use of long-
lived SCOT columns and standards permits this approach. In the case of im-
portant or difficult situations, or in research applications, the analyst
should also use additional techniques. Section 9 described further techni-
ques. Additional identification of an environmental oil sample with a
source oil can be achieved by the method of artificial weathering described
at the end of Section 6. The changes in oils that were artificially wea-
thered were a 15% drop in the n-C.. 7/n-C.. and pristane/phytane ratios, and
alterations of fine structure in the portions of the chromatogram below
the elution of n-C R. Appendix C provides further information on the wea-
thering of petroleum oils.
37
-------
SECTION 9
RESULTS AND DISCUSSION
The results of correlating the samples described in Section 1 and
Appendix G are presented in Table,6. For each spill sample, all 16 source
samples were considered. Over 10 possible sets of correlations exist, and
correct identification by chance was improbable. A definite and correct
correlation was made in 74% of the cases. Only one probable correlation was
incorrect. EPA had prepared three variants of a sample without supplying
the corresponding source oil. The analysis indicated that none of the given
sources was likely for these samples.
TABLE 6. CORRELATION OF 17 SOURCE OILS3 AND 35 SIMULATED SPILLSb BY
GAS CHROMATOGRAPHY
Correlations made by Woods Hole
Result
26 Unique correlations
6 "Probable" correlations (a second
source possible)
3 Indefinite or probable correlations
All correct
1 Errord
No correlation exists"
a Eight crudes, five fuels, and four re-refined (freshwater, saltwater,
and beach sand, with or without one of four spill-control chemicals
added).
b Prepared by EPA with 1-, 3-, and 30-day exposures.
c See Appendix H.
d Error attributable to presence of oil-based spill-control chemical.
e The source of these samples was not supplied to Woods Hole by EPA.
Since this comparison of sources and simulated spills iiv/olved a v~~
riety of oils and weathering conditions within the range encountered in
actual water quality management, the method has an excellent chance of cor-
relating real spills with their sources. In particular, no incorrect "def-
inite" correlations were made. Table 7 breaks down the pe"..'.-'manea of the
38
-------
correlation method by the variables involved. With the range of situa-
tions covered, the success rate (spill sample corresponds to one and only
one weathered possible source) does not vary significantly and is about 70
to 80%.
TABLE 7. PERCENT SUCCESS OF GAS CHROMATOGRAPHIC CORRELATION OF
"SPILLS" AND "SOURCES"
a
Category % Success
Crudes 86
Products 73
1 to 3 day exposure 80
30-day exposure 72
No additives 75
With additives 73
Overall 81
a Percent Success = percent of spill samples correlated with one and only
one source sample. See Table 6. "Spills" supplied without corresponding
source oils not considered. Success = 26/32. All correlations made were
correct.
The data in Tables 6 and 7 strongly suggest that the method may be a
useful tool in water quality management. However, some alternative combi-
nations of the 17 source oils in Table 6 would have considerably lowered
the success rate, and some would have enhanced it. The excellent perfor-
mance of the method in this trial experiment cannot be expected to hold for
all cases. The success rate could have been much lower if EPA had chosen
some other sources to generate several weathered spill samples. Comparison
of each spill sample with 16 possible sources (when there were actually 17
possible sources) is a much more complex situation than many real environ-
mental situations present. The fewer the possible sources, the simpler cor-
relation is, and generally, the higher the success rate.
The second test of the method was made by obtaining representative sam-
ples of the oils occurring in various geographical areas (described in Sec-
tion 1) and analyzing them. Table 7 clearly shows that mild weathering (ex-
posure for up to 30 days) has minimal effects on the ability of the method
to correlate oils. Therefore, success in uniquely correlating oil spills
aiter brief weathering depends mainly on distinguishing all possible sources
from one another. *
39
-------
The performance of an oil analysis method in a given geographic loca-
tion can be estimated by determining how many of the possible sources in the
area can be distinguished from one another, practical performance will tend
to be limited by the presence of similar oils, and not by short-term wea-
thering.
To reduce the number of required analyses to manageable proportions,
we estimated statistically whether some oils could be distinguished from
others. This was accomplished by comparing the peak height ratio RSD's of
the possible source oils and the environmentally weathered samples. These
estimates cover only a few cases, and the following results are close to
those that would be achieved by further analyses of all possible pairs.
Samples were obtained from the cargoes of 22 tankers in the Port of Portland,
Maine. Twenty-two samples of crude oils were analyzed. The percentage of
oils distinguished from each other are listed below.
Oils distinguished from:
All other oils 27
All but 1 other oil 50
All but two other oils 13
All but 3 other oils 5
All but 4 other oils 5
The crude oils arriving in Portland could not be uniquely distinguished
in a majority of the cases. In practice, tankers arriving from the same oil
shipment port bear virtually indistinguishable (by this method) cargoes.
Nevertheless, even if all arrivals over 2 full weeks were considered, one-
quarter of the cargoes were unique, and one-half could only arise from one
of two ships. For samples that are not unique, a useful narrowing of the
field is possible. The same type of model correlation was applied to the 30
petroleum samples from Greater New York Harbor (Table 8).
The model correlations in Greater New York Harbor are excellent. All
samples can be classified according to type by the analysis. All samples
are uniquely identifiable, except for some No. 2 fuel oils. The correlation
for No. 2 fuel oils was also high 50% correlated uniquely, and most of the
pairs were collected on widely separate locations or at different times.
However, unweathered oil samples were not being matched with weathered oils.
All samples were fresh.
ANCILLARY TECHNIQUES
Though a GC method often yields conclusive results in oil correlation
problems, cases often arise that require additional confirmatory evidence for
distinguishing an oil pair. The information gained from gas chromatography
can assist in solving these problems by use of additional gas chromatographic
techniques.
40
-------
TABLE 8. ABILITY OF GAS CHROMATOGRAPHY TO DISTINGUISH AMONG
30 OILS AND OIL PRODUCTS3 FROM GREATER NEW YORK HARBOR
Product Type
No. of samples
Results
Gasoline
No. 2 fuel oil
12
14
Not analyzed
7 unique, 3 pairs indistin-
guishable oils, 1 trio of
indistinguishable oils
No.
No.
4 fuel oil
6 fuel oil
4
8
All unique
All unique
Miscellaneous :
Kerosene
Marine diesel
Nigerian crude
Asphalt
Overall
1
1
1
1
Unique
Unique
Unique
Unique
77% unique, 20%
able pairs, 3 %
able triplets
indistinguish-
indistinguish-
a Each oil was analyzed and compared with the 29 others to determine dis-
tinguishability.
41
-------
High resolution gas chromatography is one possible approach. The use
of a second detector, such as a sulfur-sensitive flame photometric or elec-
tron capture detector, would yield more information on specific parameters
without additional time or effort. The SCOT OV-101 column is ideally suited
(good resolution, low bleed, low gas flow), for GC-MS instrumentation.
Mass spectra may have distinguishing features not clear in the flame ioni-
zation detector (FID) trace alone. Low voltage MS or GC-MS is undoubtedly
the best way to demonstrate unequivocally the presence of oil-derived aro-
matic hydrocarbons in organisms. The capability of gas chromatography may
also be extended by applying preliminary separations before analysis. Sep-
aration of aromatics (several groups) and saturates by adsorption chroma-
tography or separation by molecular shape (clathrates, molecular sieves)
before gas chromatography are proven techniques. Preparative gas chroma-
tography on packed column may also be used to obtain fractions for analysis
by other methods.
A second way of utilizing GC in conjunction with other techniques is to
take advantage of the information gained by GC. The number of samples to
be analyzed by expensive or complex techniques can be drastically reduced
by using GC analysis as a very high performance screening procedure. De-
tailed examination of samples by high-resolution or low-voltage mass spec-
troscopy, by neutron activation analysis, or by various isotopic measure-
ments may only be economically feasible with the use of screening to reduce
sample load. The neutron activation method (Lukens et al., 1971) may be
particularly promising and complementary to GC. The trace elements analyzed
may enable one to distinguish identical oils shipped in different vessels,
pipelines, etc., by virtue of the pickup of trace element fingerprinting.
The boiling-point distribution indicated by GC indicates some informa-
tion useful in ruling out nonproductive techniques. Fuel oils wii] often
tend toward legal sulfur content limits and sulfur analysis is not \.rom"
in distinguishing them. Nickel and Vanadium contents of lower-boiling c
tilled products are very low. Estimates of the extent of evaporative wea-
thering by GC may help in evaluating data from many other methods, sn-:'. as
infrared, ultraviolet, and fluorescence measurements. The similarity o " the
boiling point distribution curve to that of some known product type is use-
ful in estimating the possible sources of oil contaminants in a given geo-
graphic area.
PRACTICABILITY
Many methods have been proposed for correlating environmental oil pol-
utants with their sources. A complex, expensive, slow activation analysis
scheme is claimed (Lukens et al., 1971) to have extremely high discrimin-
ating power, and extensive geochemical analysis of samples has been sug-
gested (Koons et al., 1972). Both of these methods are undoubtedly of some
value. Both invest a large amount of time and effort in each sample, or
use highly specialized and expensive equipment.
By contrast, the gas chromatographic method is less expensive and can
be performed by a competent chemical technician with some experience in gas
42
-------
chromatography. Analyses of bulk oil samples require about 15 minutes of
technician time and 1 hour of instrument time. This flexibility makes GC
analysis useful for spill-source correlations, monitoring environmental oil
levels routinely, and research involving the environmental fate of oil.
The research potential of gas chromatographic analysis has been suc-
cessfully demonstrated in a study of the fate of an oil spill in West
Falmouth, Massachusetts, for over 2 years. The location, concentrations,
and compositional alteration of No. 2 fuel oil in marine and marsh sedi-
ments and in organisms were determined and correlated with biological data
(Blumer et al., 1970, 1972; Sanders et al., 1972). The gas chromatographic
techniques presented in this Manual are as reproducible and of higher reso-
lution than those required for the West Falmouth study.
A second advantage of gas chromatographic analysis is that the re-
sulting chromatogram can guide the selection of applicable techniques for
more detailed analysis. Similar oils might be recognized as fuel oils,
which are blended to meet sulfur content requirements.
A third advantage of gas chromatographic analysis is that the sample
is essentially separated according to the evaporating tendency of its con-
stituents. Since evaporation is the most active environmental force alter-
ing oil composition (especially on a short-term basis), this separation
clearly indicates evaporation effects and indicates the unaltered portion
of an oil fingerprint.
Several disadvantages traditionally associated with the gas chromato-
graphic analysis of oil pollutants have been overcome in this method. The
Western Oil and Gas Association concluded (Kreider, 1971) that it was not
possible to compare oil samples except on the same instrument, using the
same column, on the same day. We have been able to remove these limitations
for all but the closest comparisons. The same group has objected that rel-
atively high resolution columns, as used in this study, have very limited
lifetimes when subjected to repeated oil sample analyses. However, the
columns used in our study have shown a useful life of at least 200 analyses.
43
-------
REFERENCES
American Society for Testing and Materials 1975. Tentative Method of Test
for Preparation of Sample for Identification of Waterborne Oils. ASTM
Method No. D3326-74T. J_n 1975 Annual Book of ASTM Standards, Part 31,
pp. 561-564.
Blumer, M., Sass, J., Souza, G., Sanders, H., Grassle, F. and G. Hampson
1970. The West Falmouth Oil Spill. Woods Hole Oceanographic Insti-
tution Technical Report No. 70-44. Unpublished Manuscript.
Blumer, M., Ehrhardt, M. and J. H. Jones 1972. The Environmental Fate of
Stranded Crude Oil. Proceedings, U. S. Navy Oil Pollution Conference,
Washington, D.C.
Blumer, M. and J. Sass 1972a. The West Falmouth Oil Spill. Data available
in November, 1971, II. Chemistry, Woods Hole Oceanographic Institution
Technical Report No. 72-19. Unpublished Manuscript.
Blumer, M. and J. Sass 1972b. Indigenous and Petroleum-Derived Hydrocarbons
in a Polluted Sediment. Mar. Pollut. Bull. 3:92-94.
Blumer, M. and J. Sass 1972c. Presistence and Degradation of Spilled Fuel
Oil. Science, 176:1120-1122.
Ehrhardt, M. and M. Blumer 1972. The Source Identification of Marine Hydro-
carbons by Gas Chromatography and Spectrometry. Environmental Pollu-
tion, 3:179-194.
Farrington, J. W., Teal, J. M., Quinn, J. G. and K. Bruns 1973. Intercal-
ibration of Analyses of Recently Biosynthesized Hydrocarbons and Pe-
troleum Hydrocarbons in Marine Lipids. Bull, of Environ. Contam. &
Toxicol. 10(3):129-136.
Koons, C. B., Monaghan, P. H. and G. S. Blayliss 1972. Pitfalls in Oil
Spill Characterization: Needs for Multiple Parameter Approach and
Direct Comparison of Spill Materials with Specific Parent Oils. Un-
published Manuscript, cited in: Industrial Research, January 1972.
Kreider, R. E. 1971. Identification of Oil Leaks and Spills. 1971 Proc.
Joint Conf. Prevent. Contr. Oil Spills., Washington, D.C., API, pp 119-
124.
44
-------
Lukens, H. R., Bryan, D., Hiatt, N. A. and H. L. Schlesinger 1971. Devel-
opment of Nuclear Analytical Techniques for Oil-Slick Identification,
Phase IIA. Final Report. Gulf Energy & Environmental Systems Co.,
Gulf Radiation Techn., San Diego, Calif., Report No. GULF-RT-A-10684,
Contract No. AT(04-3)-167.
Meinschein, W. G. 1969. Hydrocarbons - Saturated, Unsaturated and Aromatic.
In Organic Geochemistry, Ed. by G. Eglinton & M. T. Murphy, Springer-
Verlag, N.Y., pp330-356.
Sanders, H. L., Grassle, J. F. and G. R. Hampson 1972. The West Falmouth Oil
Spill, I. Biology., Woods Hole Oceanographic Institution Technical
Report No. 72-20. Unpublished Manuscript.
45
-------
BIBLIOGRAPHY
Adlard, R. E. 1972. A Review of the Methods for the Identification of Per-
sistent Hydrocarbon Pollutants on Seas and Beaches. J. Inst. Petrol.,
58:63-74.
Adlard, R. E., L. F. Greaser, and P. H. Mathers. 1972. Identification of
Hydrocarbon Pollutants on Seas and Beaches by Gas Chromatography.
Anal. Chem., 44:64-73.
Blumer, M. 1957. Removal of Elemental Sulfur from Hydrocarbon Fractions.
Anal. Chem., 29:1039-1041.
Blumer, M., and W. D. Snyder. 1965. Isoprenoid Hydrocarbons in Recent
Sediments: Presence of Pristane and Probable Absence of Phytane.
Science 150:1588-1589.
Burns, K. A., and J. M. Teal. 1971. Hydrocarbon Incorporation into the
Salt Marsh Ecosystem from the West Falmouth Oil Spill. Technical
Report No. 71-69. Woods Hole Oceanographic Institution, Woods Hole,
Massachusetts. Unpublished Manuscript.
Clark, R. C., and M. Blumer. 1967. Distribution of Paraffins in Marine
Organisms and Sediments. Limnol. & Oceanogr., 12:79.
Cole, R. D. 1971. Recognition of Crude Oils by Capillary Gas Chromato-
graphy. Nature, 233:546-548.
Davis, J. B. 1967. Petroleum Microbiology. Elsevier Publishing Co.,
Amsterdam, London, N.Y.
Ehrhardt, M. 1972. Petroleum Hydrocarbons in Oysters from Galveston Bay.
Environmental Pollution, 3:257-271.
Eisma, E., and J. W. Jurg. 1969. Fundamental Aspects of the Generation of
Petroleum, jn Organic Geochemistry, Ed. by G. Eglinton & M. T. Murphy,
Springer-Verlag, N.Y., pp. 676-696.
Ettre, L. S., Purcell, J. E., and S. D. Norem. 1965. Support-Coated Open
Tubular Columns, 1. J. Gas Chromatog., 3:181-185.
Grob, K., and G. Grob. 1969. Splitless injection on Capillary Columns,
Part 1. The Basic Technique; Steroid Analysis as an Example. J.
Chrom. Sci., 7:584-591.
46
-------
Mattson, J. S. 1971. "Fingerprinting" of Oil by Infrared Spectrometry.
Anal. Chem., 43:1872-1873.
McNair, H. M. and E. J. Bonelli. 1969. Basic Gas Chromatography. Chapter
III. Varian Aerograph, Walnut Creek, California.
Purcell, J. E., and L. S. Ettre. 1966. Support-Coated Open Tubular Columns,
II. Applications in Trace Analysis. J. Gas Chromatog., 4:23-27.
Robinson, C. J. 1971. Low-Resolution Mass Spectrometric Determination of
Aromatics and Saturates in Petroleum Fractions. Anal. Chem., 43:1425-
1434.
Robinson, W. E. 1969. Kerogen of the Green River Formation. In Organic
Geochemistry, Ed. by G. Eglinton & M. T. Murphy, Springer-Verlag,
N.Y., pp. 619-636.
Rogers, M. A. N. J. Bailey, and C. R. Evans. 1971. A plea for Inclusion
of Basic Sample Information when Reporting Geochemical Analyses of
Crude Oils. Geochem. Cosmochim., 35:622-636.
Smith, H. M. 1968. Qualitative and Quantitative Aspects of Crude Oil
Composition. U. S. Dept. of the Interior, Bureau of Mines Bulletin
642.
Speers, G. C. and E. V. Whitehead. 1969. Crude Petroleu. In Organic
Geochemistry, Ed. by G. Eglinton & M. T. Murphy, Springer-Verlag, N.Y.,
pp. 638-675.
Straughan, D. 1972. Factors Causing Environmental Changes After an Oil
Spill. J. Petrol. Tech., March 1972: 250-254.
Welte, D. H. 1969. Organic Matter in Sediments. In Organic Geochemistry,
Ed. by G. Eglinton & M. T. Murphy, Springer-Verlag, N.Y.
47
-------
APPENDICES
APPENDIX A PETROLEUM CHEMISTRY
Many excellent articles and books discuss the origin and chemistry of
petroleum. Individuals with major responsibility in the area of oil pollu-
tion should be familiar with this literature. This section presents a
brief summary of the main features of oil chemistry and geochemistry. For
the purpose of this manual, "oil" or "petroleum" is a carbon-rich liquid
that can be made to flow from the earth. This definition excludes natural
gas and pit-mined asphalt.
The major constituents of oil span the entire molecular weight (MW)
range from small molecules (He, CH.) to very high molecular weight polymers
and polymer aggregates. Nevertheless, the low molecular weight compounds
(MW below ca. 2,000; "maltenes") are usually distinguished from the higher
molecular weight polymeric material ("asphaltenes").
The predominant element in both these classes is carbon. The high
molecular weight polymer is rich in linked aromatic structures, including
heteroaromatic compounds of N, S, and 0, and metal-organic complexes.
Little is known of the structural composition of the polymer fraction. The
forces between individual polymer molecules are strong, so that much of ' >c-
high MW fraction is essentially a single, giant molecule. Analysis of t
fraction involves its destruction in varying degrees.
In contrast, the lower molecular weight fraction is an extremely com-
plex mixture of mutually dissolved carbon compounds that offer much greater
possibility for chemical analysis than the high molecular weight material.
The relative proportions of the two fractions vary greatly among oils and
also among oil products. The low molecular weight fraction responsible for
the fluidity of oils is always present (except in some tars and asphalt) in
sufficient quantity for GC analysis.
In addition to the major high and low MW components mentioned above,
oils contain many other lesser constituent classes. Organic and mineral
particles and emulsified water and its solutes are present in minor and
highly variable amounts.
The lower molecular weight fraction of oil is a complex mixture of com-
pounds that offers the greatest promise for correlating oil. It is also
chemically and geochemically the best understood fraction. The following
list of petroleum constituents shows some of the major types of these com-
pounds found in many oils.
48
-------
Low Molecular Weight Hydrocarbons
Saturated "NSO" Compounds Aromatic
Straight chain Mercaptans 1-6+ rings
Isoprenoid Thiophenes Alkylated
Branched chain Aryl thiophenes Polyalkylated
Cyclic, polycyclic Indoles, N heterocycles Cycloalkylated
Carboxylic Acids Metal-organic complexes
Etc.
High Molecular Weight "Asphaltenes"
Structural Units
Chains
Crosslinks
Multi ring aromatics
Heterocycles
Metals
Most of these compounds are mutually soluble, stable, and relatively water-
insoluble substances. Their evaporating tendencies vary greatly with molec-
ular size. Some components are permanent gases at normal temperatures (me-
thane, ethane), and others are high melting solids (longer-chain, normal
paraffins).
Gas chromatography partially separates the low molecular weight oil
fractions according to their individual evaporating tendencies. Individual
substances elute from the column and are detected in an order closely re-
lated to their boiling points. The gas chromatograms yields information
about a major fraction of an oil by separating it according to a property
(volatility) that is predominant in environmental weathering (evaporation)
and in refining (distillation) of oils. Gas chromatograms of oils are useful
in gaining an understanding of oil and oil compositions. Appendix E contains
illustrations of gas chromatograms of a variety of crude oils and oil pro-
ducts. All gas chromatograms have been determined under comparable condi-
tions. In all cases, the vertical axis is the detector response (closely
proportional to the amount of carbon-bearing material eluted), and the hori-
zontal axis represents the time elapsed during the temperature-programmed
run. Text Figure 3 indicates the position of some identified peaks on a
chromatogram of a standard No. 2 fuel oil. The tick marks on chromatograms
indicate the expected peak position for normal heptadecane (n-C-,7). The
baseline signal is determined separately and is subtracted from the sample
signal to determine the contribution of the oil to the signal.
Even casual inspection of the gas chromatograms of crude oils in Appen-
dix E reveals certain basic features, with great variety in the relative
contributions of these features. The oils all show a signal greater than the
baseline over wide ranges of elution temperature. This signal is caused by
the continuous elution of an extremely complex overlapping of peaks. Each
49
-------
substance is in very small amounts. The sum of the peaks as sensed by the
detector is a relatively smooth curve with minor variations in intensity.
A particularly good example of this feature is given by Bachaquero crude
oil (Figure 4), in which this characteristic accounts for almost the entire
signal.
The complexity of this mixture of compounds should not be underesti-
mated. It is typical of crude oils generally. Present columns cannot per-
form a complete separation. The best capillary columns are capable of re-
solving up to about 100 compounds in the interval between two successively
eluting normal hydrocarbons. However, it is known from oil geochemistry
that thousands of substances occur in the C-,,. region..
All oil analyses are partial analyses. In the case of gas chromato-
graphy, this incompleteness can be seen in the chromatogram itself. The
"unresolved envelope" is not an experimental artifact or an indication of
poor technique; it is a consequence of oil composition.
The gas chromatograms of many crude oils show a series of well-resolved
peaks in addition to the complex mixture of substances just discussed.
Chemical fractionation of oils shows that the generally prominent series of
peaks (Chromatograms 7, 16, 20, 30 etc.) is due to the presence of an ex-
tended homologous series of normal straight chain hydrocarbons. No other
prominent homologues yielding such an extended series of discrete peaks has
ever been found in any crude oil. The size of these peaks relative to each
other and to the unresolved envelope varies from oil to oil, and also
according to the column resolution.
In addition to these discrete normal hydrocarbon peaks, several other
relatively large peaks occur in many crude oils, particularly those re-
sulting from the isoprenoid hydrocarbons pristane and phytane (Text Figure
3). The origin of these substances is discussed later. They occur at dif-
ferent levels in different oils and are relatively unaffected by weathering.
The broad envelope, normal peaks, and pristane and phytane peaks
account for the appearance of most of the crude oil chromatograms in Appen-
dix D. They are the major features compared in the correlation of oil sam-
ples. Other peaks of intermediate size occur in many oil chromatograms
and can be used for oil correlation as well. There is no absolute distinc-
tion between the partly resolved, smaller peaks and the larger ones. In
practice however, oils with different patterns of small peaks generally have
different patterns for the larger peaks. The smaller features therefore
serve more as a confirmatory check of a postulated correlation than as a
starting point.
It is necessary to clarify some terms to avoid confusion. The peaks in
an oil chromatogram will frequently be called "pristane peak" or simply
"pristane," or "n-C1ft" (meaning normal octadecane). This nomenclature is a
highly useful shorthand. The peaks named in this manner occur at the same
location in an oil chromatogram as do the separately analyzed authentic sub-
stances after which they are named. The authentic substances have been
50
-------
_J
o
LU
O
o:
o
o
tr
LU
o
CD
09
o
S
O
o
09
o-
re
o
re
CO
»*-
o
E
re
2
re
o
6
O>
Ll.
51
-------
isolated from many oils. However, the presence of peaks at these expected
locations does not prove the structure of the substance found. The peaks
in an oil chromatogram do not represent pure substances. The signal is a
sum including a substantial contribution from the unresolved envelope. In
the case of pristane and phytane, several closely related isomers may also
be eluting together. It is not possible to make an exact separation of the
signal into peak and background components without considerable chemical
fractionation. Correlations simply identify the presence or absence of dif-
ferences. Knowing the chemical identity of major contributing components
of peaks is important in assessing the effects of weathering.
The main features and comments that have been made about crude oils
apply, with slight modification, to many oil products as well. Many fuel
and bunker oils (Appendix D) are made by the distillation of crude oils,
along with additions of other materials in small quantities. These products
do not necessarily resemble any particular crude oil. Many crudes may
become admixed in the manufacturing processes. These products will gener-
ally have a restricted boiling range corresponding to the distillation range
used in manufacturing. The No. 2 fuel oils are an example of this. No. 6
fuel oils are higher-boiling, or residual materials. Number 4 fuel oils are
often a blend of Nos. 6 and 2 (Chromatograms 44, 46, 48, 49). The general
shape of the above-background envelope of an oil sample (especially the
higher ends are not subject to evaporation) may reveal whether it is a dis-
tillate or a blend of distillates and residues rather than a crude oil. The
relative amounts of individual normal paraffins and pristane and phytane
present, compared to the unresolved complex mixture, depend on the oils used
in manufacture and are not typical of the product. Some products are "de-
waxed" or otherwise treated in ways that remove normal paraffins or satu-
rated compounds (for example, lubricant oil; See Chromatogram 51). It has
been argued that a sample could not be a No. 6 fuel oil because No. 6 has a
prominent series of peaks and the sample in question did not. Though this
sample undoubtedly differed, the argument is incorrect.
Not all oil products that are referred to by different names are nece-
ssareily chemically different in a clear-cut way. Marine diesil and No. 2
fuel oil (Chromatograms 46 and 47) differ in boiling range, but we did not
have enough samples of marine diesel to be sure that the distinction was
always clear. Materials such as gasoline, asphalt, and mineral oil have
been chemically altered, and their chromatograms do not have any strong re-
semblance to crude oils. Automobile gasoline is quite light and could be
analyzed under different conditions than those useful for most other oil
products.
Though the gas chromatograms of different oils from different fields
are relatively constant and fixed by geological considerations, the output
of refineries varies with economics and technology. In practice, routine
sampling programs in a given area should reveal marked changes and enable
water quality managers to take them into account.
52
-------
APPENDIX B PETROLEUM GEOCHEMISTRY
Petroleum geochemistry provides a framework for understanding the
main compositional features of oil, the differences between oils and other
naturally occurring hydrocarbons, and the effects of several common wea-
thering processes on oil composition. The question of the origin of oil is
vigorously debated. The information presented here is a simplification of
widely held theories or well-established facts.
The ultimate source of the carbon in oil is living matter, but the
components of oil are not simply the individual structures present in
living matter, concentrated by natural processes. The central issues of
oil geochemistry concern the extent, location, and manner in which nonhydro-
carbon matter is transformed to petroleum hydrocarbons, and the extent to
which oil is siray-ly a collection of small amounts of materials indigenous
to life (by geological processes). Much more is known about the origin of
the fraction of MW below 1,000, especially the hydrocarbons, than about
other constituents.
Plants, sediments and soils, and organisms from many environments have
been analyzed for low molecular weight hydrocarbons. The only abundant
constituents of oil also found abundantly distributed at the earth's surface
are the normal hydrocarbons, pristane, and some of the singly branched hydro-
carbons. The normal hydrocarbons from C to C,~ are also present in oils;
adjacent members are present in similar quantities. This distribution con-
trasts markedly with that of plants, organisms, clean recent sediments, and
shales that contain a limited range (mostly Clt- to C_,-) of normal hydrocar-
bons and a strong predominance of odd-numbered over even-numbered compounds.
The total natural hydrocarbon content of recent sediments is often 10 to
100 ppm. Thus the hydrocarbon distribution and frequent presence of ele-
vated hydrocarbon levels can distinguish background hydrocarbons from pe-
troleum contaminants. This and related points have been discussed in detail
by Blumer et al., (1972 a,b,c).
Part of the normal hydrocarbons and pristane in oils might arise di-
rectly from surface materials, but much of the n-hydrocarbon content must
arise from other matter by subsurface processes. Heat, pressure, and cata-
lytic action of clays and other matter and bacteria operating over geologic
time lead to extensive chemical alterations. This process, diagenesis, de-
stroys many original components and creates many new ones. The exceedingly
complex chemistry gives rise to the extremely complex mixture of low-level
components, the unresolved envelope of oils. This mixture is not (with rare
and doubtful exceptions) found in plants and animals, whose biochemistry
directs the synthesis of a relatively few materials with high specificity.
An example of diagenetic alteration is the fate of chlorophyll, an extremely
abundant biological molecule. The side chain, phytyl alcohol, would yield
phytane if it were reductively removed from chlorophyll by reactions easily
accomplished in the laboratory. A different sequence of reactions involving
the loss of one carbon atom could also easily yield pristane. These com-
pounds in oil arise in part from plant-produced chlorophyll by a series of
reactions as suggested by Figure 5. The relative and absolute amounts of
53
-------
these two substances produced and preserved depend on many factors. These
factors vary in a complex manner with source material, age and depth of
burial, geology of the matrix, temperature, and subsequent history of oil
in the reservoir.
The complex mixture of aromatic hydrocarbons found in petroleum has
never been found as a natural component in organisms, though some bacteria
apparently do synthesize a few simple, unsubstituted or mono-substituted
aromatic compounds (Meinschein, 1969). The presence of a complex mixture
of aromatic hydrocarbons (as shown by column chromatography followed by
GC-MS) is probably the best evidence that material is oil-derived. This in-
formation may indicate that material is oil-derived even in cases too com-
plex to allow correlation with a source. Contaminants arising from coaltar
products may interfere, but they are rare. Since they are not formed by
organisms, complex mixtures of aromatic hydrocarbons can be interpreted as
contaminants if found in organisms.
The geochemistry of petroleum also offers some natural experiments on
water-related oil weathering processes. Evaporation, dissolution, and mi-
crobial catabolism of some components are the primary processes occurring
in weathering. Dissolution and microbial catabolism occur in certain types
of oil reservoirs. The simplest example is the "open" type of reservoir,
in which oil is trapped at a high point in a continuous, permeable stratum
bounded by relatively impermeable strata. Oil in such traps is held im-
mobile with respect to groundwater, which flows through strata and past the
immobile oil. The oil is water washed and the soluble components are se-
lectively removed. The groundwaters may also bring in oxidizing agents
(oxygen, sulfate) and bacteria, and so subject the oil to the action of bac-
teria. Such oils are frequently missing most of the usually abundant low-
molecular weight compounds (below G1fJj and they frequently have relatively
low contents of the normal hydrocarbons. Sometimes pristane and phytane are
not prominent in these oils. Many detailed experiments and field observa-
tions on such oil fields have demonstrated that the loss of low-molecular
weight materials is due to their generally higher water solubility. The low
content of normal hydrocarbons and often of isoprenoids is due to the selec-
tive oxidation of these compounds by bacteria. Laboratory experiments show
that the normal hydrocarbons are oxidized much more rapidly than isoprenoids
and other classes of compounds.
APPENDIX C PETROLEUM WEATHERING
The primary processes affecting oil in water environments are evapora-
tion of the more volatile constituents, dissolution of the more soluble com-
ponents, and preferential oxidation by microorganisms of the straight-chain
and slightly branched hydrocarbons. These same processes can affect the
composition of oils before they are ever produced.
Evaporation is the major and fastest-acting oil alteration process in
most spills. Within hours, compounds up to about ten carbons are heavily
depleted, and floating oil loses most compounds up to about 13 carbon atoms
in a few days. Beached oils may evaporate more rapidly. Oils exposed by
54
-------
CHLOROPHYLL
PHYTYL GROUP,
REDUCTION
LOSS OF 1C ATOM
AND REDUCTION
PHYTANE
PRISTANE
Figure 5. A Geochemical Source of Pristane and Phytane.
55
-------
the EPA (Text Tables 6 and 7) partially lost compounds up to n-C after 1
month floating on water outdoors, but n-C and above were little affected.
Evaporation eventually removed most of the components boiling below n-C
or n-Cn,,. This required approximately 1 year.
Laboratory-simulated evaporation of suspected source oils paralleled
the effect of exposure on spill samples. A serious alteration of the most
useful analytical ratios did not occur in samples fewer than 30 days old.
Evaporation may make it difficult to classify a sample as a crude which lost
light compounds, compared to an oil product such as No. 4 fuel oil, in which
these compounds were removed in manufacture.
In our experience, short-term (weeks) dissolution has negligible influ-
ence on the higher-boiling, nonevaporating fractions of oil used in this
correlation method. It probably resembles .evaporation enough to be confused
with it.
The most erratic weathering factor is microbial alteration of the oil,
with preferential oxidation of normal hydrocarbons and, in extreme cases, of
the isoprenoids. This oxidation can affect most of the useful analytical
ratios used in correlating oils.
For short-term (30 days or less) exposure of oils, there is little evi-
dence that microbial oxidation occurs to any important extent. It has been
repeatedly shown in the laboratory that under ideal conditions, oil-oxidis-
ing bacteria can alter a sample markedly in a few days. The marine envir-
onment seldom, if ever, provides such ideal conditions. Howevvr, several
proposed oil-spill control and cleanup methods enhance the rate of such re-
actions. Seeding with bacteria and nutrients or addition of hay will pro-
bably accelerate bacterial action markedly.
Older oil samples may show varying degrees of n-hydrocarbon loss
through bacterial oxidation. These samples may be correlated by the use of
other parameters.
The effect of oil weathering seems to be small on correlation ability
for oil-rich samples that are not very old. Oil recovered from sediments
and/or organisms, or after long periods of exposure, has probably been al-
tered by all weathering processes. It is still possible in many rases to
identify the material as oil and to place constraints on the initial compo-
sition (For example, The West Falmouth oil spill). However, correlating a-
bility in the sense demonstrated in Test Table 6 will often be severely
degraded.
A very rough line of division may be drawn between oil-rich water,
slicks, sediments, and tarballs only a few weeks old (for which the chances
of correlation with a suspected source are excellent) and oils that are older
or more heavily diluted and admixed with indigenous hydrocarbons in organisms
and sediments. In these cases, identification of the specific ,-ource will
frequently be tenuous (though if the spill source is known previously, a con-
tinuous sequence of changes consistent with the known geochemistry of oil may
56
-------
easily be observed). Indigenous hydrocarbons in organisms and in sediments
may be distinguished from typical oils. A strong odd over even predominance
of n-alkanes and a relatively simple component mixture are typical of indig-
enous hydrocarbons. Odd/even ratios near 1:1 and the presence of a complex
unresolved mixture is typical of petroleum-derived material.
A more severe problem in interpretation may arise if the sediments or
organisms already contain oil residues from previous chronic pollution, or
if spill and baseline samples are not available. In general, the mixed ori-
gin of such materials may be indicated by the sample chemistry, and useful
data for monitoring damage or recovery can be collected. Correlation will
require careful research.
APPENDIX D SAMPLE CHROMATOGRAMS
TABLE D-l. CHROMATOGRAPHY COLUMN, INJECTION NUMBER AND MISCELLANEOUS
PARAMETERS USED TO OBTAIN THE SAMPLE CHROMATOGRAMS OF APPENDIX D.
Chromatogram Column Injection
No. No. No.
Description
Column baselines:
1
2
218
218
218
Standard No. 2
fuel oil:
4 216
5 218
6 219
Short-term
weathering study:
7 219
8 219
9 219
Spill control
chemicals:
10 218
11 218
12 218
13 219
C-17, C-18,
pristane and
phytane:
14 219
15 219
74
132
148
124
149
181
57
61
60
77
123
122
62
111
112
So. Louisiana crude
So. Louisiana crudeweathered 24 h.
So. Louisiana crudeweathered 1 mo.
Magnus dispersant
W. G. Smith surface collecting agent
Shell herder
No. 6 fuel oil with shell herder
Mixture only
Mixture with Bachaquero crude
(continued)
57
-------
TABLE D-l. (continued)
Chromatogram Coltimn Injection
No. No. No.
Description
Carrier gas
effects study
(No. 2 .FO, STD):
16
17
18
Crudes :
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Petroleum
products :
44
45
46
47
48
49
50
51
Dexsil 300"
Column :
52
219
219
219
219
219
219
219
219
219
219
219
219
219
219
219
219
219
216
216
219
219
218
218
219
219
219
219
219
218
218
218
218
219
219
218
218
220
207
209
210
57
158
161
145
174
163
160
144
155
164
146
168
170
173
156
158
77
94
62
22
197
199
200
35
202
152
157
149
158
178
183
159
24
3
With helium
With nitrogen
With nitrogen and reduced flow rate
So. Louisiana
Lago medio
Lagc mar
Agha Jari
Lago treco
Ceuta
T. J. light
Nigerian
Nigerian /P.O.
Guanipa
Orito
Zarkum
Qatar
Kuwait
T.J. 102
Murban
La Rosa
Lagunillas (Lago)
Arabian light
Chevron ventura crude
Miss. - lime
Penn. - strawn, Stevens Co. Reg.
Penn. - strawn, Eastland Co. Reg.
Bachaquero
Ordovician-Ellenburger
Gasoline
Kerosene
No. 2 fuel oil
Marine diesel
No. 4 fuel oil
No. 2 fuel oil
Asphalt
Crankcase oil (used)
No. 2 Fuel oil standard
58
-------
TABLE D-2. DATA FOR OIL AND OIL PRODUCTS EXEMPLIFIED IN APPENDIX D
Signal intensity ratios
Chromatogratn Column
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
No.
218
218
218
216
218
219
219
219
219
218
218
218
219
219
219
219
219
219
219
219
219
219
219
219
219
219
219
219
219
219
219
219
216
216
219
219
218
218
219
219
219
219
219
218
Pris/
phyt.
___
1.83
1.81
1.88
1.61
1.67
1.61
1.35
1.37
1.83
1.61
0.87
0.86
0.96
1.01
0.98
0.95
2.03
2.00
1.98
1.14
0.85
0.88
0.82
1
0.94
0.96
1.00
0.94
1.29
1.29
1.36
1.36
0.85
1.20
n-Ciy/
pris.
__
1.29
1.41
1.41
1.34
1.31
1.29
2.02
1.54
1.41 '
1.34
2.06
1.96
1.94
1.88
1.78
1.86
1.28
0.85
1.07
1.30
3.67
3.41
2.95
1
2.58
1.34
1.01
1.93
0.84
1.38
1.21
1.04
0.96
1.55
n-Clg/
phy.
__
1.65
1.90
1.90
1.68
1.65
1.67
2.04
2.07
1.89
1.68
1.54
1.52
1.57
1.68
1.58
1.57
2.02
1.40
1.79
1.19
2.85
2.60
2.08
1
2.23
1.18
1.08
1.58
0.98
1.51
1.35
1.16
0.99
1.61
n-C /
/"» '
___
1.44
1.34
1.40
1.28
1.31
1.25
1.33
1.02
1.37
1.28
1.16
1.12
1.19
1.12
1.12
1.12
1.28
1.22
1.17
1.24
1.10
1.15
1.15
1
1.09
1.09
0.94
1.15
1.10
1.18
1.22
1.18
0.84
1.15
n-C /
B.C.'
_._
3.42
4.36
4.34
4.97
5.10
5.19
5.25
3.50
4.01
4.97
5.31
4.90
5.77
4.42
4.08
4.44
5.79
3.26
3.84
5.14
6.68
6.38
5.87
1
2.73
1.48
5.90
5.23
3.98
4.67
3.81
1.02
5.33
Comment
d
d
d
b
b
b
c
a
a
a
d
d
a
d
d
a
e
e
c
c
c
c
c
c
a
c
a
c
a
a
a
c
a
c
c
c
c
a
a
a
a
a
a
d
59
-------
TABLE D-2. (continued)
Signal intensity ratios
Chromatogram Column
No.
45
46
47
48
49
50
51
52
No.
218
218
218
219
219
218
218
220
Pris/
phyt.
___
1.81
0.87
2.06
1.41
n-C /
pris.
___
1.41
1.37
1.39
1.40
n-Clg/
phy.
___
1.90
1.29
2.04
2.07
n-C /
n~C18
1.34
0.92
1.41
0.95
n-C /
B.GT7
4.36
2.86
7.41
5.83
Comment
d
b
a
a
a
d
d
e
a Values reported are for single determination.
b Values reported are averages of 11 to 13 determinations.
c Values reported are averages of two to five determinations.
d Samples were not oils or oil products containing any of the desired
components.
e Poor resolution.
60
-------
, J. 1_,,
,____[ _
>
n
-4_j_,
j.... 4- i -
~l-'
4_.:zL ._|";i4_;ji-t--
. ., . .. t I
......... __ gj. T_ , ,_ji
-. .-(_ ' - : i
. _J ,. lr_
- _1. .- ,l-_ ,--
^ g .. ^
t-
61
-------
62
-------
63
-------
116!
-f- i i ' ]-*(--
_l±jJL_L ..JJ J-
64
-------
18
-------
21
66
-------
67
-------
1/wir
' i i ! I ' - r - t : i
_., _ . <_ _ . _..,j,._. (_ _ . i,i,_.i
. ; r. f . f r i | _ fTpi.. LL
_.i i, L...JL. i.,..' f ' .; L i. i r i i j r j
30
68
-------
-------
til
it III
*,*» f it
it«gi«*ii» r
.__L
70
-------
- ! : ! I : I
... * t . 1 - « J,.- I* L,.- 8 ,'
' 1J
71
-------
-------
73
-------
APPENDIX E INTERCALIBRATION OF ANALYSES OF RECENTLY BIOSYNTHESIZED HYDRO-
CARBONS AND PETROLEUM HYDROCARBONS IN MARINE LIPIDS
This Appendix briefly describes intercalibrating analyses of recently
biosynthesized hydrocarbons and petroleum hydrocarbons in marine lipids. A
detailed report of this intercalibration exercise was reported by Farring-
ton, et al (1973).
Evaluation of the extent and the severity of petroleum contamination of
the marine environment requires the comparison of data obtained in differ-
ent laboratories using the same or different analytical methods. We have
examined the validity of comparing data collected in our laboratories. A
cod liver extract was spiked with a distillate fraction (287° to 450°C) of
South Louisiana crude oil. Each laboratory then analyzed the sample by its
own commonly used methods. The concentrations of recently biosynthesized
and petroleum hydrocarbons per unit weight of lipid were reported. Our in-
tention was not to reproducibly and accurately identify the source of the
petroleum contamination. Indeed, we expect that most of the samples we an-
alyzed from the continental shelf and deep sea, if contaminated, would have
been contaminated by more than one source. Even if they were contaminated
by only one source, the presence of recently biosynthesized hydrocarbons,
the probability of weathering of the petroleum hydrocarbons before incorpor-
ation into the sample, and the probability of not having suspect oil samples
would make it extremely difficult to identify the source.
The general method employed in each laboratory was to saponify the lipid
extract, isolate the nonsaponifiable lipids, and then isolate the hydrocar-
bons from the nonsaponifiable lipids using thin layer or column chromato-
graphy on 7.5 ft. or 10 ft. 3% Apiezon L on Chromasorb W columns with the
column temperature programmed from 80 to 100 C to 280 C at 5 or 6 C per min-
ute. Quantities of hydrocarbons were determined by comparing the detector
signal from the sample hydrocarbons, or by comparing the sample signal with
the signal of a known quantity of an n-alkane analyzed on the same day using
the same procedures.
The results are reported in Table E-l and demonstrate that our interla-
boratory precision is sufficient to allow valid comparison of analyses of
hydrocarbons in the boiling range. The results also demonstrate that for
the same boiling range, each laboratory gave a fairly good estimate of the
level of petroleum hydrocarbon contamination. The precision and accuracy
would vary somewhat depending on the composition of the petroleum hydrocar-
bon mixture with respect to the relative concentration of the nonpolar n-al-
kanes and naphthenes, and the more polar mono-, di-, and poly-aromatic hy-
drocarbons. Further intercalibration analyses using a variety of petroleum
products such as paraffinic crude oils, naphthenic crude oils, fuel oils,
bunker oils, and lubricating oilsall spiked in various matrices such as
water, sediment, lipid-rich and lipid-poor organismsare needed to demon-
strate the validity of interlaboratory comparisons of data.
74
-------
o
I-H
PM
M
1 1
Pi
W
M
Q
O
CJ
60
co
£5
0
9
^Q
O
C>
Pi
Q
ttl
M
m
I
w
o
Q
H
v^
CO
w
CO
§
3
0
<§
o
PS
Q
ad
fa
0
£3
0
H
PQ
M
S;
C5
Pi
w
H
25
rH
1
W
w
hJ
3
H
CO
Q
en
.Pul
PS
Q
cu
60
CO
o
r*>.
r-x
CO
"d"
CO
CO
CO
in vo CM
CM rH
cfl »H
r^
4-1 TJ
co o
w o
cfl
75
-------
APPENDIX F PROCEDURE FOR CLEANING VARIAN CAPILLARY INJECTOR
CAUTION: It is crucial not to allow particles to be flushed into the
line leading to the column!
1. Turn off column oven and injector heaters and allow to cool several
hours. Shut off air and hydrogen.
2. Remove the column and the instrument side panel. Disconnect the
carrier gas supply line and the split exit line at the closest accessible
point to the injector: be sure the proper lines are located and that no
valves or flow controllers are between the disconnect line ends and the in-
jector. Remove septum nut, septum, and glass liner.
3. Step 2 makes accessible four openings into the injector unit: col-
umn connection, carrier gas inlet, injection port, and split exit. Plug the
gas inlet and split exit lines and connect a tight-fitting piece of teflon
or polypropylene tubing to the column fitting; the other end should fit a
syringe tip. Be sure the tubing and plugs used contain no soluble plasti-
cizers. Flush several 5 cc portions of particle-free methylene chloride
through the column connector and catch in a flask; inspect each washing for
color and for fine glass particles.
4. Disconnect the tubing and tightly plug the column opening with a
blind stainless steel plug furnished with the column. Flush solvent through
the carrier gas line and out the injection port.
5. Cap the port tightly, using a piece of teflon or polyethylene be-
hind the septum, and flush several rinsings of CH-C1 through the carrier
gas line and out the split exit line. This line is particularly dirty and
should only be flushed in this direction.
6. Dry injector with gas stream, reassemble connections in original
fashion, check all reconnected fittings for leaks with soap solution, re-
install liner and septum, and reheat with carrier gas flowing and column
exit unplugged. When fully heated, reconnect column and reset gas flow
rates if necessary.
76
-------
CO
M
M
W
UJ
F
CO
W
H
f%
hM
Prf
O
CO Pu
Pd
M
nj
6
0
J^
f"{
u
r
i-
CJ
i
c
-v^.
r
r-
u
i
e
--^
O
r~
u
c
r
r
cj
e
.
CO
H
Pw
0
CO
C!
H
4-1
H
C
O
O
i-H
H
a
CO
cfl
CO
1 )
1 1
rl
a
CO
^
o
eq
- 00
H rH
CJ
a
0
H4J
r-»
l±~ 1
CM
*
HCO
H
ew
.
4-1
>,
("H
A- i
P-1
CJN
1 1 1 1 i I 1 oo crs I 1 1 I 1 1 I 1 1 r^~ i l CNI i i vo 1
Illl-l 1 1 1 1 1 1 1 1 1 ro | itNilini
1 1 1 1 r^ 1 1 1 1 1 1 1 1 1 1 1 1 II 1
VO O C^J O r*^ CTv OO CT^ O O CO CTt LO -3" U"} r^ rH C^ ON C^ CTv to CO 00 ^"1
ij-icocNi co n m in «d~ *
sOinCTNin| ^j- QQ Q^ , | \cj Q\ ipj r*^ r^*. csi co **o to **o oo r^ rH vo CN ^J" oo **o
rHO-c|-vO^D^Dr^^OLO^Dr~-vDvOvOvOCN|Ou-lC^
rOmmCNi-HrHrHrHrHrHrHrHrHrHrHr-lrHrHO i-HrHrHrHrHO
O 00 CTN vDvOi rcNOror ro O LO vo i 1 vO
cOoo rHONLOrHOOv
rHrHi-HrHrHr-HrHt-HrHrHrHrHrHrHrHrHrHrHO rHrHi-HrHt-HO
LO CO CO tO i 1 i 1 to *^O CNl r*^ CNI i I <] u"i OO to 00 Sf ^J" **O CNJ c\l *^t" CO to
-d- <)-C^rHrHOCT<
COCOCOCNt li IrHrHi li 1' li li IrHi lOOOi 1 i IOi li li IO
CU
^rj
CU 3
O O 4-1 T3 rH >-l
CO CU -PrfO -43 3'rlO
O --dco KO< OcrfffioO S-ioo
IS 3 O O JEJ ^ CO " "H O ^*~x pr]
M^: t£ SUCOrH " rHOCO
*cJ » * 'COOJl-jcU C5
-H[2: -cd t!S *!3:tS -£S -CU-H ^JOH-ii-J
O ^ ^ fd ^ * ^5 ^2 * £2 ^ ^E [2 *T3 O ^H J-i W ^^ / ^2
^cd'Pn-'fopLi'fe*- 'd OO CM
i i pMC/3-i-fw feco cn coc/Drcnj-iiHvr)cn co OTC/D
QJ** CO * «v * «s UUO 'Hcd'Cd
3 W ** ** *rH **CO ^^COCO **CO »*n^ % ^) 'V-l^O t~H *
i-l ) ^j ^p} rf^ ^ tf^ ^i ,j^ rc^ ^ ^ _r^ ^| (f*j t r^ r(^j pj i44 O CJ ^3 CU ^"H ^H tf^
^JJ4-»O-U^4J4-I334J34->4JH4JCO jSCd'-^H -H4-I
CN O 0 Cl hJ pJOdCJOOCOCCQJC *H a>cucu
-------
T3
cu
4-1
d
o
o
w
lj
i
CU M
O O
0 4J
cu cd
M 6
cu o
<4H J-4
CU XI
& o
to
o o
I
C P-l
r-. oo
rH rH
O CJ
C i
co
rH4.
o ?
cd
C,
60
H
co
r-
rHCO
O -H
I rl
cj p I
CO 4-1
M X!
PL. P-.
cd
CO
M C
o o
cd T3
a
cu o
cxo
H rH
rH
rH -rl
H &
O CO
cu co
O rH
rl rH
3 -H
O &
CO CO
IIICOIrHllllimilllll
III I I I I I I I I I I I I
I
o
o\o\OrHrH i o\o\cr>ocr>oooooo cooooo
O O rH rH rH I OOOrHOrHrHrHrHrHrH OOO
o\ oo vo oo -d" vor~-cj\r>.LOoooocNOcTv ooincrv
CTV CT* CN CT* 00 I "^ ""3" CO CO rH O
S CU rH
3 CO 0) CU
O -H
o
o
M
CO
OO
3 MH cd -H tS
O O
& ts cu
cO C u-i ^
COCOlHO)
CU
C rH
O }-i 3
Pn en Pn 3
fe MCOCOCOCOCOO
o
2 &
Pn
&
Pt!
(3 C
co o
J-l O CU v£>
CO*»COCO ^i*.^*.*
34J3 3C04J4J4J4J4J
OCOOQGCCCPI
CU CO 'CO
cr 3 4J 3
cd o C o
H-H 3
ex & o
CU (!)
o o
M rl
cu cu
o o
M rl
cu cu cu
OrHrHrHrH OrHrHrHrHrH OrHrHrH
J-lrHrHrHrH J-lrHrHrHrHrH J-lrHrHrH
3 -H -H -H -H 3 -H -H -H -H -H 3 -H -H -H
o ex ex ex ex o ex ex ex ex ex o ex ex ex
cococococococococococococococococococococo
60 I
C cu
H }-i
cd J-i
cu ->
4-1 CU
cd >
M *H
3 4J
Tj O
(U
cu pi
xl co
H CU
CU p!
oo a)
cd oo
cd
60
C 60
4J -rl
O 4J
cu u
rH CU
rH rH
O rH
O O
o
cu
o cu
cd o
M-l Cd
cu cd co 3 M
co co 3
= cd ^ co
T3 JS ^
cu & cu i-i
rH O M iH
3 -H cu 6
B T3 TJ BC CO
H CU (3
CO 4J -H rH
Cd rH C
<) O CO Q)
P-i -H C XI
Pd Td o co ts
C -H I I
XI -H 4J I I
o cd W co
cd co -H co O
~~ ' > !3
W
78
-------
^-^
T3
a
p
r
4-1
C
0
o
,
&
ti
PC
E-i
t
s
X-N
CU
CO
X-N Cfl
CU 43
CO
cfl M
43 CU
4-1
rH Ctf
. 1 f>
H 3
O N '
s^
-M
j t r*
+J M
C CO
ctf CO
ca s-i
t-i CU
CU Cu
Cu ca
CO vH
H 13
a
4-1
CO -H
3 X
C cu
60 M
ctf O
S o
1 1
O Pi
^8
*
CO
C!
O
rl
jl
+J
Ctf
13
H
0
CU
4-1
cu
TJ
!-i
3
O
<4H
O
4-1
O
&
4-1
M-l
O
cu
00
CO
M
CU
>
cfl
»^
i^%
iH
rH
CO
fj
w
cu
C
cu
60
cu
n
CO
c
cu
>
H
60
CO
O
H
4J
,3
,0
^
CO
C cfl
B i-i
^ 1
iJ ^>
rH -H
o a
0 O
60
>. ctf
& *1
T3 S-<
CU O
J-J M^
0
CU T3
»-i CU
)-i 4J
O Cu
0 CU
o
CU X
ft CU
c
ctf T3
O 0)
4-1
4-J O
ca cu
JS M
4-1 H
o co
cu a TH
0 43
C 4J 4J
CU O
M C W
cu o
IH cu cu
M (-1 rH
CU CU MH
4-1 £ CU
c n
H HI
co C
C cu
ctf 43 >
4-> -rl
Q) -rl 60
> &
Ctf CO
60 T3 O
0) -H
HI 4-1 4J
CO Cfl Cfl
CU !J
i t,
r~1 H
CO 4-> CU
0 43
H CO 4J
CU iH T3
43 -H C
a a. cfl
CO
rH
O < v
M EC
4-1 >-, CO
C 43
O Cu «
0 cfl t2
M CO
rH 60
rH O >
H 4-1 43
Cu cfl 4J
cfl 6 C
0 O
CU M B
43 43
H 0 rH
O
7,9
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-77-163
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Correlation of Oils and Oil Products by Gas
Chromatography
5. REPORT DATE
August 1977 issuing date
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
0. Zafiriou, M. Blumer, J. Myers, D. Stainken*
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Woods Hole Oceanographic Institution
Woods Hole, Mass. 12543
*Environmental Protection Agency, Edison, NJ
10. PROGRAM ELEMENT NO.
1BB610
11. CONTRACT/GRANT NO.
Grant No. 15080HEC
12. SPONSORING AGENCY NAME AND ADDRESS
Industrial Environmental Research Lab.
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
- Cin., OH
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/12
15. SUPPLEMENTARY NOTES
16. ABSTRACT
A gas chromatographic method is presented for identifying the discharge
source of petroleum oil pollutants found as slicks and shoreline residues.
The method is used to match environmentally altered oils with unweathered
source samples, a technique known as "fingerprinting." Analyses of arti-
ficially aged oils and of potential spill sources found in Greater New York
Harbor and Portland, Maine, indicated a high rate of success for the method
in realistic situations.
The method was demonstrated to be suitable for routine use v/Jt;h
weathered and unweathered samples, and for monitoring levels of hydroc^L
bons in organisms and sediments. The method can be modified to stud" the
fate and effects of lower levels of petroleum hydrocarbons.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Gas Chromatography
Crude Oil
Degradation
Identifying
Chromatographic Analysis
Chemical Analysis
b.IDENTIFIERS/OPEN ENDED TERMS
F ing er pr in t ing
c. COSAT! Field/Group
07C
11H
13. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS {ThisReport)
Unclassified
21. NO. OF PAGES
90
20. SECURITY CLASS (This page}
Unclassified
22 PRICE
EPA Form 2220-1 (9-73)
80
iUS GOVERNMENT PRINTING OFFICE 1977757-056/6519
------- |