R.; Tejada, S.; Bumgarner, J.; Duffield, F.; Waters, M.; Simmon, V. F.; Hare, C.; Rodriguez, C.; Snow, L. Washington, D.C., Sept
(14) Gerasimenko, Y. E.; Schevchuk, I. N. Zh. Org. Khim. 1968,4, 2198. (15) Konieczny, M.; Harvey, R. G. J . Org. Chem. 1979,44, 2158. (16) Cormack, M.; Spielman, M. A. Org. React. (N.Y.)1946,3, 83. (17) Newman, M. S. J. Org. Chem. 1944,9, 518. (18) Gold, A. Anal. Chem. 1975,47, 1469. (19) Eisenstadt,E.;Gold,A.Proc. Natl. Acad. Sci. U.S.A. 1978,75, 1667. (20) Gold, A.; Schultz, J.; Eisenstadt, E. Tetrahedron Lett. 1978, 4491. (21) Ittah, Y.; Jerima, D. M. Tetrahedron Lett. 1978,4495. (22) Ruehle, P. H.; Fischer, D. L.; Wiley, J. C., Jr. J . Chem. Soc., Chem. Commun. 1979,302.
1978, EPA-60019-28-027. (11) Ames, B. N.; McCann, J.; Yamasaki, E. Mutat. Res. 1975,31, 347. (12) Cook, J. W.; Hewett, C. L. J . Chem. SOC. 1933,401. (13) Vollmann, H.; Becker, H.; Corell, M.; Streeck,H. Liebigs Ann. Chem. 1937,531, 1.
Received for review April 25,1980. Accepted July 21,1980. This work was supported in part by a Biomedical Research Support Grant, National Institutes of Health, in part by a Faculty Research Grant, University of California, Berkeley, and by the Northern California Occupational Health Center.
in Man"; Mohr, U., Schmahl, D., Tomalis, L., Eds.; IARC Publication No. 16, Lyon, France, 1977. (6) McCann, J.; Ames, B. N. Proc. Natl. Acad. Sci. U.S.A. 1976,73,
950. (7) McMahon, R. E.; Cline, J. C.; Thompson, C. Z. Cancer Res. 1979, 39, 682. (8) Wang, Y. Y.; Rappaport,S. M.; Sawyer, R. F.; Talcott, R. E.; Wei, E. T. Cancer Lett. (Shannon, Irel.) 1978,5, 39. (9) Wei, E. T.; Wang, Y. Y.; Rappaport, S. M. J . Air Pollut. Control Assoc. 1980,30, 267. (10) Huisingh, J.; Bradow, R.; Jungers, R.; Claxton, L.; Zweidinger,
Dispersion and Weathering of Chemically Treated Crude Oils on the Ocean Clayton D. McAuliffe" Chevron Oil Field Research Company, La Habra, California 90631
Jaret C. Johnson and Stephen H. Greene JBF Scientific Corporation, Wilmington, Massachusetts 01887
Gerard P. Canevari Exxon Research and Engineering Company, Florham Park, New Jersey 09732
Thomas D. Sear1 Exxon Research and Engineering Company, Linden, New Jersey 07036
Four research oil spills of Murban and La Rose crude oils were made off New Jersey. Two slicks were immediately sprayed with a dispersant; two, after 2 h. Average oil contents by IR analysis of a CC14 extract of water samples collected 30-90 min under immediately treated slicks a t 1, 3,6, and 9 m were 0.7,0.7,0.3, and 0.2 mg/L for La Rosa and 3.1,2.4,0.5, and 0.4 for Murban. The highest concentrations were 3 mg/L for La Rose and 18 mg/L for Murban. Oil concentrations for dispersion delayed 2 h were 11.1mg/L, slightly higher than found under untreated oil sampled immediately after discharge. The dispersed oil weathered very rapidly with evaporation of C1-Clo hydrocarbons greatly exceeding solution. Dissolved hydrocarbons were not found a t the method detection limit of 0.01 pg/L. The measured C1-Clo hydrocarbons were residual in dispersed oil droplets, and their sum did not exceed 50 M/L. Introduction
Four research crude oil spills discharged on the open ocean were chemically treated with a dispersant. The underlying water was then analyzed to determine (a) the dispersion of oil into the water column and (b) the rate of loss (weathering) of low-molecular-weight hydrocarbons from the dispersed oil. These tests were conducted in a manner similar to those for untreated spills conducted in 1975 (1,2). The 1975 untreated oil studies showed relatively low initial concentrations of nonvolatile hydrocarbons in the water column under the slicks (generally less than 1 mg/L) that decreased to background values in 1-2 h. Those samples containing naturally dispersed oil showed very rapid weathering. The Cl-Clo hydrocarbons detected were residual in the oil droplets, and truly dissolved 0013-936X/80/0914-1509$01.00/0 @ 1980 American
hydrocarbons were apparently not present at detection limits of a few ng/L (1,2).Samples of oil collected over time from the surface slicks showed slower weathering. Chemical dispersion is thought to accelerate the natural weathering processes. This would result in higher concentrations of oil penetrating to greater depths, and accelerated escape of volatile hydrocarbons to the atmosphere. The mechanism for this behavior was expected to be the mixing of dispersed droplets having high specific surface areas in near-surface water, causing rapid loss of volatile hydrocarbons. An untreated slick, although constantly exposed to the atmosphere, may be less susceptible to evaporation than dispersed oil because its lower surface-to-volume ratio tends to retard transport (by diffusion) of volatile hydrocarbons. Oil emulsified in water is removed from most of the wind's influence, so that it does not travel as far as a surface slick. This minimizes the possibility of oil stranding or entering biologically sensitive areas. Review and discussion of the alteration of oil on a water surface provide greater details of the processes (3). Experimental Methods
General Operations. In November 1978, four spills were conducted -40 km off New Jersey and 95 km south of Long Island, New York. Each spill was 1.67 m3 of one of two crude oils (Murban from Abu Dhabi and La Rosa from Venezuela). Each spill was discharged from a 1.9-m3tank mounted on the research vessel through two 7.6-cm hoses. The ends of the hoses were on plywood floats, causing the oil to discharge horizontally on the water surface. This minimized both evaporation losses due to discharge above the water, and
Chemical Society
Volume
14, Number 12, December 1980
1509
Table 1. General Experimental Summary date of spill time of spill time of dispersion spill location latitude longitude conditions wave height (m) wind (m/s)
(knots) air temp (‘C) water temp (‘C)
Murban 1
La Rosa 1
La Rosa 2
Murban 2
Nov 2, 1978
Nov 3, 1978
Nov 9, 1978
Nov 9, 1978
11:53 13:50
10:14 12:oo
10:19 10:28
14:04 14:ll
40’ 09’ 09” N 73’ 30’ 39” W
40’ 09’ 12” N 73’ 33‘ 40” W
40’ 09’ 18” N 73’ 32’ 00” W
40’ 09’30’’ N 73’ 34’ 45” w
0.3- 1 .O 4.0-5.5 8-1 1 15-20 14
0.3-1.0 4.0-5.5 8-1 1 15-20 14
0.3-1 .O 2.5-6.0 5-12 12-14 13
0.3-1.0 2.5-6.0 5-12 14-17 13
Wind Direction
,---*--/
I
Sample Depths
1 & 3 m - All Stations 6& 9mI at Stations3 & 0
4 \
Figure 1. Schematic of immediately dispersed oil slick and location of sample stations for typical 10-station sample run.
vertical descent of the oil into the water. The less viscous Murban (0.83 specific gravity, 39” API), discharged in -3 min; the La Rosa (0.91 specific gravity, 23.9’ API), in 6 min. The oils were treated by aerially spraying a self-mix dispersant from a pod and spray booms mounted above the skids of a helicopter. The mean droplet diameter was -1 mm. The helicopter flew in a rapid back-and-forth pattern crosswind. Spraying was from 10 to 15 m above the water surface. One slick of each oil was treated immediately with 150 L of dispersant, and one each after 2 h with 360 L. In all cases, there was overspraying (outside the slick) and a percentage loss due to wind drift. The major experimental conditions are summarized in Table I. The sea state was about the same for all spills; air and water temperatures were slightly lower November 9 compared with November 2 and 3. Sample Collection. The sampling program was designed to obtain water samples at approximately equally spaced stations on transects through the surface slicks and emulsion plumes as shown in Figure 1. Samples were taken at 1- and 3-m depths at all 10 stations, and at 6 and 9 m at stations 3 and 8. Surface samples were taken, with a small bucket, at all stations during sampling runs through dispersed oil. A sampling run took -45 min. For the immediately dispersed slicks, the first run was started a few minutes after dispersion, and the second after -1.3 h. For the two delayed dispersion tests, one sampling run was made before dispersion (untreated oil), and two after, immediately and after -1 h. For all of the spills a few samples were also taken 2-4 h after dispersion, a t station 3. The subsurface samples were collected with small submersible pumps discharging through polypropylene tubing, 1510
Environmental Science & Technology
a t -4 L/min. The pumps were attached -0.5 m below a floating 114-L steel drum towed 3 m lateral to the bow of the research vessel. In this position, the ship’s bow wave did not cause water mixing a t the sample inlets along a line suspending a 23-kg weight from the bottom of the float. The sample gear was lowered and removed from the water outside the observed slicks to avoid surface oil contamination. Two types of samples were collected: one 1.5-L sample in a 1.9-L flint glass jug, and duplicate completely filled 300-mL “soft drink” bottles (new) with crown caps. The 1.9-L jugs were cleaned by rinsing three times with carbon tetrachloride (cc14)that was checked for purity by infrared (IR) spectroscopy. Immediately after collection, 50 mL of this cc14 was added to each jug from an all-glass dispensing pipet. The jugs were sealed with Teflon-lined metal screw caps and hand-shaken for -10 s to initiate the solvent extraction of the dispersed oil. The CC14 also prevented bacterial degradation of the hydrocarbons. In the laboratory, the samples were shaken 2 min to complete the extraction. Before 300-mL sample collection, about 30 mg of mercuric chloride (HgC12) was added to each bottle to prevent hydrocarbon biodegradation. Each bottle was then flushed with reactor-grade helium and sealed with a crown cap (polyvinyl chloride seal). At the time of sample collection, each bottle was uncapped, filled to within 3 mm of the top, and resealed with a new crown cap. The small air space minimized loss of volatile hydrocarbons to this gas space and minimized possible contamination of sample by hydrocarbons (as well as CC14 vapors) that may have been in the atmosphere during sample collection. Samples of each crude oil were taken from the spill tank in glass bottles with Teflon-lined screw caps. In the laboratory these oil samples were equilibrated with sea water collected outside the spill area, to provide equilibrium dissolved hydrocarbon concentrations in sea water. Aerial Control and Photography. A small twin-engine high-wing aircraft served as a control platform from which to direct the dispersant-spraying helicopter and to direct the research vessel to each sampling station. Periodic color photographic runs were made over each slick by using a vertically mounted camera in the floor of the aircraft to provide photographic documentation of the oil slicks and their chemical dispersion. Chemical Analysis. Total extractable organic matter was measured on the single 50-mL portion of CC14, with a Fourier transform IR instrument, as absorbancy at 2930 cm-’. This method measures other CC14-soluble compounds such as organic acids, esters, and alcohols in addition to the crude oil. The C C 4 extracts of a few samples were further analyzed for total nonvolatile (C,4+) hydrocarbons, by removing polar
.25 19.1 10.8
4.06
.18
0 1
3
6
9-
9
.IO
6 b 6
Stations Timeafter t15 +20 +23 I dispersion ( m i n ) L
+30
& J
+32
I
I
6 +48
7
t52
- - - - - - - - -- - - - - - - - - - - - - - ~~
1 n tejsect)
9 +61
10 +66
Total extractable organic matter (ppm) in water samples collected during first sample run through La Rosa crude oil spill immediately dispersed (oil spilled 10:19, dispersed 10:28-10:35). Figure 2.
organic compounds with a silica gel column and reanalysis by IR ( 4 ) . Volatile hydrocarbons (C1-Clo fraction) in the water samples were analyzed by a gas equilibrium method ( 5 ) .Forty milliliters of Murban and La Rosa oil samples were equilibrated with 140 mL of sea water by hand shaking gently and periodically for 24 h or more. This water was filtered (from one 50-mL glass syringe into a second) to remove separate-phase oil that may have been dispersed during oil-water mixing. Twenty-five milliliters of this water was gas equilibrated five times. These successive analyses were used to measure the equilibrium concentrations of individual CI-Clo hydrocarbons for the two crude oils and to calculate individual hydrocarbon distribution coefficients ( 5 ) . The water samples from the various stations and depths were analyzed with a single equilibration using the measured distribution coefficients to calculate concentrations. This gives sufficient accuracy and saves time and cost of multiple equilibrations. Method details are given in ref 1 and 5 .
Results and Discussion Visual and Photographic Observations. When dispersant was applied to fresh La Rosa, no sudden change was apparent. However, in time this oil became a thin sheen, as contrasted with the thick, black, asphaltic appearance of the thicker oil in the downwind leading edge portion of untreated La Rosa crude oil. Murban crude oil changed dramatically when dispersant was immediately applied. A distinct whitish-brown subsurface plume appeared quickly. Over several hours, this plume dispersed in the water column, grew in area, and diminished in color and visibility. A thin-transparent surface oil sheen gradually appeared during this time period as some of the emulsion droplets resurfaced and broke. Application of dispersant after 2 h of weathering appeared to have little effect on Murban crude oil, based on visual and photographic observations. Dispersal of weathered La Rosa crude oil did appear effective. However, some oil reappeared within 10-15 min after dispersant application. These visual observations gave qualitative indication of the dispersion of crude oils by chemical treatment, but chemical analyses are needed for quantitative interpretation. Oil Dispersion as Determined by Infrared Analysis. The large number of chemical analyses prevents a complete tabulation of the results for total extractable organic matter (OM). Some of the analyses will be presented in graphical form to document chemical dispersion. As expected, the highest concentrations in the water column were attained after immediate dispersion as compared with dispersion after 2 h.
.11
OB Stations
Time after dirpenion !mini +75
t78
I Stations
+m
+83
+86
b
I I
b
Time after 195 dspenion (mini
0
+99
(& t103
!I"t.rsect1
I
h1W m
ill0
t118
Figure 3. Total extractable organic matter in water samples collected during second sample run through La Rosa crude oil spill immediately
dispersed. Interpretation will concentrate on immediate dispersion results. The crossed transects of a sampling run (Figure 1)permit a three-dimensional analysis of plumes of dispersed oil (Le., in crossed vertical planes). However, the limited sampling at 6 and 9 m (only at stations 3 and 8) may give a distorted (narrow) view of dispersion a t these depths. On the basis of natural dispersion of oil into the water column (2) and the lack of chemically dispersed oil a t 6 m ( 6 ) ,significant dispersion of oil down to 6 and 9 m was not expected. These depths were added only to verify the prior results. Figure 2 shows the extractable OM concentrations with depth along the two transects of the first sampling run following the immediate dispersion of La Rosa crude oil (the vertical scale exaggeration is -45X). The contour for 0.25 ppm was a t -9 m at its deepest point; for the 1.0-ppm contour, 4-5 m. The shape of the 2.0-ppm contour on the transect of stations 6-10 is interesting in its asymmetry. Relying as it does on one data point for its symmetric shape, this might be suspected as an experimental artifact. However, the second set of transects, -1 h later (Figure 3),produced the same type of contour. Figure 3 also shows the lower concentrations brought about by dilution of the plume in a larger volume of water. Petroleum hydrocarbons (C14+) were determined on three of the extracts. Extractable OM was 2.24, 1.25, and 2.54 mg/L; Volume 14, Number 12, December 1980
1511
Total Extractable Organic Matter. m r n
Total Extractable Organic Matter, ppm ..
C C
3
.o
1
-- 3 s E
8 a
Murban Depth profile 81 min after dispersion
6
9 i 111111
I
I Illlill
i
I 1 Ill111
Figure 4. Comparison of concentration-depth profiles at one station for various times under the immediately dispersed La Rosa crude oil spill.
1
31
20.5136023.2
.09
$17
I
I Stations
1
Timeafter 118 dispersion. lminl
2
3
4
i21 +?3 +29 L
5 t32
- --- - - - -
Ilntenectl
Figure 5. Total extractable organic matter (pprn) in water samples collected during first sample run through immediately dispersed Murban The dashed crude oil spill (oil spilled 14:04,dispersed 14:ll-14:16). contour for 4 ppm is based on the station 1-5 transect.
C14+ hydrocarbons were respectively 1.43, 0.72, and 1.97 mg/L. Petroleum hydrocarbons averaged -75% of the total extractable OM. This is in the range previously observed for a much larger number of analyses (7). However, the actual crude oil content of the original CCld extracts is higher because hydrocarbons 5 pprn). The polar organic compounds removed by silica gel appear to exceed the extractable OM from background water samples outside the oil spill areas (particularly noticeable when the extractable OM ranges from 0.2 to 1ppm). A possible explanation is that crude oil acts as an organic solvent, extracting and concentrating natural organic compounds in sea water. Another way to r-iew the dilution is shown in Figure 4 for the immediately dispersed La Rosa spill. Concentrations a t station 3 in the center of the plume are plotted with depth over time. Because each depth concentration is a single analysis, great reliance should not be placed on an individual point. As expected, a steady decrease in concentrations toward background values occurred over time. Concentration lines for the immediately dispersed Murban spill (Figure 5) show higher values than with La Rosa (Figure 2). Dispersed oil was also found in higher concentrations a t greater depths (almost 1ppm a t 9 m). 1512
Environmental Science & Technology
I I IIIII
I
I I111111
I
I I Illlll
Figure 6. Comparison of concentration-depth profiles for La Rosa and Murban crude oils at about the same time following discharge and dispersion.
Figure 6 compares concentration-depth profiles of the two crude oils, for samples from the center of the plume a t similar times after oil discharge and dispersion. Again, each concentration is a single data point. A rough material-balance calculation indicates that the Murban crude oil was almost completely dispersed, whereas the La Rosa was about half dispersed. These evaluations concur with visual impressions of effectiveness. The data for the two spills that were allowed to weather for 2 h before dispersion do not allow such clear graphical display. Most values for total extractable OM were much lower than those from the immediately dispersed spills. One explanation is the larger area to be treated after 2 h, with a consequently larger water volume available to dilute an equivalent amount of oil. Most of the oil was concentrated in the leading (downwind) part of the slick, in perhaps only 10%of the total slick area, as observed by Hollinger and Mennella (8). The dispersant was applied uniformly over the whole slick rather than concentrated on the area of heavy oil. Therefore the dispersant-to-oil application rate to the heavy oil portion of the slick was not as high as for the immediately dispersed spills, which were treated before appreciable spreading had occurred. Weathering also would have increased oil viscosities, and thereby would have decreased dispersant effectiveness. A summary of the total extractable OM in water under the four research oil spills is shown in Table 11. It includes only values exceeding 0.10 ppm (approximately two times background). Untreated oil dispersed naturally in the water to a lesser extent than chemically treated oil. Immediate dispersion was more effective than after 2 h, but most of the difference may be attributed to differences in application rate of dispersant to the oil. The greatest difference between oils was evident when they were dispersed immediately. Murban oil concentrations were higher a t all water depths than for La Rosa. The slightly higher concentrations for La Rosa compared with Murban following delayed dispersion may reflect differences in chemical application and/or sampling locations. The oil concentrations found under the nondispersed slicks are similar to those found for these same oils in 1975 (2, 7). During one of the earlier Murban tests, when the wind and waves were higher (7-10 m/s, 0.6-1.0 m), somewhat higher oil concentrations were found (2.3 ppm a t 1.5 m, 1.0 ppm at 3 m). The air and water temperatures were the same for the Murban tests in 1975 and 1978, so that oil viscosities and densities were
Table II.Summary of Carbon Tetrachloride Extractable Organic Matter in Water under Four Research Oil Spills (Concentrations in mg/L) Murban
La Rosa depths, m
nb
max
mean
rib
max
4 3
0.22 0.51
0.13 0.26
1 2
0.95 0.16
7 7 2 1
0.23 1.05 0.65 0.29
0.15 ,027 0.38
8 4 1 1
0.18 0.11 0.14 0.12
0.13 0.10
16 14 5 1
2.24 2.96 0.50 0.25
0.69 0.67 0.31
13 9 4 4
17.80 10.20 1.oo 0.95
3.10 2.45 0.45 0.40
mean
not dispersed 1 3
0.14
dispersed after 2 h 1 3 6 9
dispersed within 10 min 1 3 6 9
Background concentrations (mglL) 1 m, 0.061;3 m, 0.050:6 m, 0.048; 9 m, 0.051.
the same. The higher oil concentrations thus appear to result from greater mixing energy. The concentrations and dilutions with times reported here for nondispersed crude oils are similar to those reported by Cormack and Nichols for 0.5 m3 of Ekofisk crude oil discharged in the North Sea (9).The dispersed oil concentrations in the upper meter of water are also similar to those obtained by these workers when oil was sprayed from a boom on a moving boat and dispersant was sprayed from a second boom positioned so that the spray covered all of the oil. Oil Weathering as Measured by CI-Clo Analysis. Infrared analysis of CCl4 extracts provides a measure of total oil in water samples but is relatively insensitive. As used in this study, the method had a limit of detection of -0.02 m g b . The method does not give information on individual hydrocarbons, classes of hydrocarbons, or degree of weathering (loss of lowmolecular-weight hydrocarbons). The gas equilibrium method ( I , 5 ) permits the measurement of most individual hydrocarbons in the Cl-ClO fraction with a limit of detection of 2 ng/L (ppt) for alkanes and cycloalkanes and 10 ppt for aromatic hydrocarbons. This analysis permits the loss of low-molecular-weight hydrocarbons to be followed with time (1-3). If adverse biological effects (immediate toxicity) result from oil spills, they are thought to be produced principally by the more soluble low-molecular-weight hydrocarbons such as benzene and toluene. Of importance, therefore, are the concentrations of the dissolved hydrocarbons and the duration of organism exposure to them. When water is equilibrated with crude oils, the C1-Clo soluble fraction comprises over 98% of the total soluble hydrocarbons ( I O ) . For typical oils, benzene plus toluene constitute 70-80% of the soluble aromatic hydrocarbons, and 62-78% of the total C,+ soluble hydrocarbons ( I O ) . Gas Chromatograms. Gas chromatograms of (a) dissolved hydrocarbons in sea water equilibrated with an excess of Murban crude oil from the spill tank and (b) C1-Clo hydrocarbons residual in dispersed oil droplets in a water sample collected under the chemically treated Murban oil spill are shown in Figure 7 . The GC column was 6 m of 3.2-mm stainless steel tubing packed with 10% UCW-98 silicone fluid on Chromosorb W-HP. The column was temperature programmed from 60 to 145 "C at 6 "C/min. A 30-cm precut (backflush) column was in a sample valve oven at 100 "C. The column was backflushed a t 4 min, which prevented >Cl0 hydrocarbons from entering
Number of samples.
the 6-m column. A 2.0-mL (1.6-mm diameter) sample loop in the sample valve oven introduced 1.5 mL (at 100 OC) of the 20-23 mL of gas flowed from the 50-mL equilibration syringe through the sample loop. The numbers over or near the individual hydrocarbon peaks are the relative retention times in hundredths of minutes. They vary slightly because of changes in column conditions, and differences in integrator start time (manual) at the time of sample injection. Each principal hydrocarbon peak has been named, and the GC amplifier attenuation is given. Figure 7A is the gas chromatogram of dissolved hydrocarbons in sea water equilibrated with Murban crude oil at attenuation of 1X (methane through pentanes) and 500 X for the remaining hydrocarbons. The partial GC (Figure 7A) is from another analysis with less attenuation, to better show the characteristic di- and trimethylbenzene peaks. Figure 7A shows the marked decrease in concentration of hydrocarbons with increase in molecular weight (carbon number), and the much greater solubility of aromatic hydrocarbons relative to the saturated hydrocarbons of the same carbon number (cycloalkanes are more soluble than alkane hydrocarbons ( 1 1 ) ) .In particular, note the large benzene and toluene peaks. The decrease is due not only to lower solubility with an increase in carbon number but also to the lower concentrations of individual hydrocarbons in crude oils (higher carbon numbers than toluene for aromatics) as carbon number increases. An increase in number of isomers occurs with an increase in carbon number. For pure hydrocarbons, normal alkane solubility decreases 6-7 orders of magnitude between carbon numbers 1 and 12. For aromatics, the solubility decreases similarly between carbon numbers 6 and 24 ( I O , 11).For example, hexane, cyclohexane, and benzene, each with six carbon atoms in the molecule, have respective solubilities of 9.5,60, and 1750 mg/L ( I I , I 2 ) . Thus benzene and cyclohexane are respectively 185 and 6 times more soluble than hexane. The aromatic-to-nalkane solubility ratio increases ( I O ) , so that dimethylnaphthalenes are over 600 times more soluble than n-Clz. Most of the gas was separated from the crude oil. Thus, the peaks in Figure 7A for methane through pentanes (particularly methane, ethane, and propane) are lower than if the crude oil was "live" (gas not removed). Figure 7B represents the CI-Clo hydrocarbons in a water sample collected a t 1 m near the center of the immediately treated Murban spill 46 min after dispersion. The attenuation is 500 times less (lo00 times for C1-C5) than in Figure 7A, and Volume 14. Number 12, December 1980 1513
0
-> 0)
X
n I
E
Trimethylbenzenes
+ 50 x
2
l o x 10-10
Attenuation
h
v1
x
500 X
10-9
Attenuation
(D N
I
Number Over Each Peak is Relative Retention Time in Hundreths of Minutes (1074 = 10.74 min) Attenuation 1 x
Alkane or Cycloalkane
Figure 7. Gas chromatograms:(A) Equilibrium concentrations of dissolved hydrocarbons in sea water mixed with an excess of Murban crude oil from the spill tank. Inset is from second chromatogram with less attenuation to show more clearly the di- and trimethylbenzenes. (B) C1-Cl0 hydrocarbons found in 1-m water sample collected 40 min after immedaite dispersion of Murban crude oil spill (total extractable organic matter was 3.8 ppm). See text for details of analytical procedures.
the peak areas (concentrations) are entirely reversed (methane through the trimethylbenzenes). This qualitatively shows not only the very low concentrations of these low-molecularweight hydrocarbons in dispersed oil (Figure 7B), but progressively greater loss with decrease in carbon number. Weathering of these low-molecular-weight hydrocarbons was very rapid. Quantitative data are presented in tables that follow. Weathering of Murban Crude Oil. Table I11 shows C1-Clo hydrocarbons in water samples from 0- to 9-m depths a t the center of the Murban slick 46 min after spraying with a self-mix dispersant. The first numerical column (oil max) gives the equilibrium concentrations of dissolved hydrocarbons in sea water that was thoroughly mixed with an excess of Murban crude oil from the spill tank. Note, as discussed above, the decrease in concentration with the increase in carbon number, and the high relative concentrations of ben1514
Environmental Science & Technology
zene and toluene. The alkane and cycloalkane hydrocarbons (>7 carbon atoms) have become so low that they are difficult to identify and separate from aromatic hydrocarbons (Figure 7). Thus n-heptane and methylcyclohexane are the highest carbon number saturate hydrocarbons shown in Table 111. In essence, only aromatic hydrocarbons were measured in solution from toluene through trimethylbenzenes. The peaks designated as alkane or cycloalkane (Figure 7B) apparently arise from the presence of separate-phase oil. Presumably these are droplets smaller than the filter used to remove most of the separate-phase oil. Shaw and Reidy (13) found the bulk of Prudhoe Bay crude oil treated with Corexit 9527 dispersant to exist as droplets smaller than 0.4 pm, but larger than 0.03 wm (determined by filtration). The concentrations of the individual hydrocarbons found in the dispersed (emulsion) plume of the Murban crude oil confirm the distribution and values indicated in Figure 7B.
Table 111. Low-Molecular-Weight Hydrocarbons in Water Samples from Various Depths Collected 46 min after Immediate Dispersion of Murban Crude Oil depth, m extractable OM, mglL hydrocarbons, sg/L
methane ethane propane isobutane n-butane isopentane npentane cyciopentane 3-methylpentane n-hexane methylcyciopentane benzene cyclohexane n-heptane methylcyclohexane to I ue ne ethylbenzene m,p-xylene o-xylene 926 trimethylbenzene 1027 trimethylbenzene 1077 trimethylbenzene 1,2,4-trirnethylbenzene 1197 trimethylbenzene total saturates total aromatics total hydrocarbons
3 2.54
1
0 11.0
3.8
9 0.95
8 0.97
(011 max) a
102 1560 2360 940 2720 870 1080 510 125 290 280 6080 270 65 140 5630 610 1550 900 68 800 370 920 300 11300 17200 28500
0.077 0.004 0.009 0.006 0.004 0.002 0.007 0.026 0.023 0.072 0.092 0.260 0.205 0.34 0.59 3.75 2.25 7.70 5.45 0.42 6.50 2.75 6.20 3.20
0.0003b 0.0004 0.0006 0.0002 0.0002 0.0006 0.005 0.018 0.025 0.033 0.004 0.076 0.52 0.42 0.067 0.37 0.50 0.61 0.62 0.8 1 0.74 0.67 1.07
1.46 38.5 40
0.070 0.002 0.004 0.003 0.002 0.002 0.005 0.012 0.006 0.022 0.027 0.095 0.066 0.067 0.135 1.40 0.80 2.55 1.85 0.12 1.55 0.65 1.40 0.66 0.42 11.1 11.5
0.000 1 0.0002 0.0003 0.000 I 0.0002 0.0005 0.002 0.005 0.008 0.010 0.002 0.024 0.103 0.096 0.025 0.13 0.16 0.21 0.18 0.19 0. 18 0.15 0.22
0.072
0.070
0.073
0.041 0.01 1 0.008 0.040 0.395 0.285 1.05 0.75 0.050 0.74 0.30 0.75 0.38
0.041 0.018 0.020 0.022 0.48 0.23 0.95 0.57 0.035 0.50 0.16 0.31 0.13
0.26 0.075 0.50 0.24 0.023 0.027 0.035 0.065 0.040
0.13 4.75 4.90
0.13 3.41 3.54
0.07 1.27 1.34
a Equilibrium concentrations of dissolved hydrocarbons when an excess of Murban crude oil from the spill tank was mixed with sea water at 25 O C . Italicized value is percent hydrocarbon found in water sample compared with equilibrium concentration of dissolved hydrocarbon (oil max). Number is relative retention time (see Figure 7).
They are very low, and the lowest carbon numbers are present in the lowest concentrations. This is the opposite of that expected if solution were an important process. Consider the hypothetical situation of oil on a water surface with (a) evaporation prevented and (b) a limited volume of water maintained in contact with the oil (Le., the laboratory conditions for mixing a sample of crude oil from the spill tank with sea water in a sealed glass bottle). Under equilibrium conditions, one would expect to find the concentrations and relative concentrations as shown in “oil max”, Table 111. Removing the restriction on water movement, but preventing evaporative loss, would result in nonequilibrium solution of hydrocarbons into the water, and the rate of solution of individual hydrocarbons would then become important (just as for evaporation). The rate of solution increases with decrease in carbon number and with class of hydrocarbon (Le., aromatic > cycloalkane > alkane for the same carbon number). Under nonequilibrium conditions, methane would go into solution faster than ethane, ethane faster than propane, etc. Similarly, benzene would go into solution faster than toluene, toluene faster than xylenes, etc. The concentration of each hydrocarbon becomes progressively lower as the degree of departure from equilibrium increases. Thus, the shorter the contact time of oil and water, the lower the concentration of each hydrocarbon in water, and the higher the relative concentrations for those hydrocarbons having the lowest carbon numbers for each class of hydrocarbons (alkane, cycloalkane, and aromatic). Because this was
not observed in the water samples under the slick, one is led to the conclusion that solution is apparently not a very important process compared with evaporation, even when crude oil is chemically dispersed. Thus, evaporation is the dominant process. The low-molecular-weight hydrocarbons that do dissolve apparently quickly evaporate to the atmosphere from nearsurface water, or dilute quickly to very low concentrations. The hydrocarbons in solution measured in the water samples (Table 111) apparently were not in true solution a t the time of collection, but residual in separate-phase oil droplets. After collection, they equilibrated between the oil droplets and water. The equilibrium solubility of C ~ Ohydrocarbons + is very low, probably less than 10 ppb. Thus, the separate oil phase in the samples ranged from -11 000 ppb in the surface sample to 940 ppb in the 9-m sample. Separate-phase oil was therefore 275-600 times the total dissolved hydrocarbon concentrations. The data in Table I11 show that the residual hydrocarbons are low in concentration, even for the surface-collected sample. Thus, the biologically toxic low-molecular-weight hydrocarbons have been quickly lost. The concentration of the least volatile, trimethylbenzene (1197), is only 1.07% of the equilibrium solubility for unweathered oil (column 1,oil max). The remaining percentages (column 3) show in general a progressive decrease with carbon number (only 0.0002-0.0006% for ethane through pentane hydrocarbons). The percentages found a t 1 m (numerical column 5 ) are even lower, showing that oil droplets a t this depth are more Volume 14, Number 12, December 1980
1515
Table IV. Low-Molecular-WeightHydrocarbons in Water Samples Collected Over Increasing Time at 1 m under the Immediately Dispersed Murban Crude Oil Spill time afler dispersion, min extractable OM, mglL hydrocarbons, p g l L
18 17.8
ethane propane isobutane n-butane isopentane n-pentane cyclopentane 3-methylpentane nhexane methylcyclopentane benzene cyclohexane nheptane methylcyclohexane toluene ethylbenzene rnpxylene o-xylene 926 trimethylbenzene 1027 trimethylbenzene 1077 trimethylbenzene 1,2,44rimethylbenzene 1197 trimethylbenzene
0.004 0.016 0.008 0.006 0.003 0.007 0.004 0.010 0.024 0.036 0.117 0.139 0.128 0.315 3.50 2.20 8.85 5.65 0.50 8.95 3.90 7.52 4.60
total saturates total aromatics total hydrocarbons
0.70 45.8 46.5
110 0.31
48
3.8
0.00038 0.0007 0.0008 0.0002 0.0003 0.0006 0.0008 0.008 0.008 0.013 0.002 0.05 1 0.196 0.225 0.062 0.36 0.57 0.63 0.73 1. 12 1.05 0.82 1.53
0.002 0.004 0.003 0.002 0.002 0.005 0.012 0.006 0.022 0.027 0.095 0.066 0.067 0.135 1.40 0.80 2.55 1.85 0.12 1.55 0.65 1.40 0.66
0.008 0.006 0.018 0.020 0.041
0.008 0.006 0.004 0.008 0.008 0.006 0.024 0.040
0.014 0.55 0.42 1.80 1.13 0.07 1.06 0.51 1.13 0.68
0.150 0.090 0.429 0.300 0.029 0.051 0.078 0.175 0.120
0.35 11.1 11.5
0.09 7.4 7.5
0.08 1.45 1.53
a Italicized value is percent hydrocarbonfound in water sample comparedwith equilibrium concentration of dissolved hydrocarbon(Table IO, oil max). Number is relative retention time (see Figure 7).
weathered than oil droplets near the surface. Accelerated weathering is noted with increasing depths as shown by decreasing concentrations and by percentages if calculated for 3, 6, and 9 m. For example, 1,2,4-trimethylbenzeneconcentrations in the dispersed oil droplets for 0, 1,3,6, and 9 m are respectively 0.67, 0.15, 0.08,0.03, and 0.007%. A part of this decrease is accounted for by the deeper samples having less dispersed oil (11ppm in surface sample + 0.95 ppm a t 9 m = 11.6). If the distribution of dissolved hydrocarbons between separate-phase oil and water are assumed to be constant, then the lower the oil content, the lower the dissolved hydrocarbon concentration in water for these low oil concentrations. The ratio for the percents for these depths are 0.67 f 0.007 = 96, confirming the greater weathering of oil in the deeper water samples. Because the samples were collected simultaneously, the accelerated weathering with increasing depth apparently relates to smaller droplet sizes. Evaporation and solution are diffusion processes; and the shorter the diffusion pathway, the higher the rate. As droplet size decreases, the surfaceto-volume ratio increases, with resulting faster loss of volatile and soluble hydrocarbons. I t seems reasonable to expect smaller droplets a t greater depth. Oil-in-water emulsions have a size distribution that ranges from 0.1 to 100 pm ( 1 4 ) .The larger droplets (some may be even larger than 100 pm (15)) will be buoyant (Murban crude oil has a specific gravity of 0.83) and will rise toward the water surface after mixing downward by wave action. However, below diameters of -2-3 pm, gravitational effects are balanced by Brownian forces. These small droplets move by 1516
Environmental Science & Technology
Brownian motion and will disperse in all directions, just as clay-sized ( < 2 pm) mineral particles stay indefinitely suspended in water. Benzene and toluene percentages in column 3, Table 111, are 0.004 and 0.067 whereas n-hexane, methylcyclopentane, cyclohexane, n- heptane, and methylcyclohexane are 0.025, 0.033, 0.076, 0.52, and 0.42, respectively. Thus benzene and toluene are very much lower than the saturate hydrocarbons as the percent sequence increases in value with increase in carbon number. This is also shown to a lesser extent for the percentages of these hydrocarbons in column 5. The low percentages for benzene and toluene also indicate that the hydrocarbons found in water samples were residual in the droplets and not in true solution a t the time of collection. These data indicate that benzene and toluene were lost more rapidly by solution than the corresponding-carbonnumber saturates and that, once lost, they were subsequently evaporated or quickly diluted. Had these aromatics been in true solution at time of collection, their concentrations should have been higher than the corresponding-carbon-number alkane and cycloalkane hydrocarbons. The concentration of methane (Table 111) is constant at -70 ppt. This reflects the expected equilibrium concentration of methane in sea water with that in the atmosphere for this region of the Atlantic Ocean (16). Table IV presents the concentrations of C1-Clo hydrocarbons in water samples collected a t 1 m over the time (18-110 min) that measurable oil was detected. These data show the rapid loss of volatile hydrocarbons with time. Even a t 18 min, the trimethylbenzenes average a little over 1%remaining in
Table V. Low-Molecular-Weight Hydrocarbons in Water Samples Collected at Two Depths and at Two Times Following Immediate Dispersion of La Rosa Crude Oil depth, m t h e after dispersion, min extractable OM, mg/L hydrocarbons, p g / L
methane ethane propane isobutane n-butane isopentane npentane cyclopentane 3-methylpentane n-hexane methylcyclopentane benzene cyclohexane n-heptane methylcyclohexane toluene ethylbenzene m,p-xylene o-xylene 296 trimethylbenzene 1027 trimethylbenzene 1077 trimethylbenzene 1,2,4-trimethyIbenzene 1197 trimethylbenzene total saturates total aromatics total hydrocarbons
0 47 4.70
3 47 2.96
0 94 1.63
3 94 1.13
(oil m a x I a
170 1740 2360 620 1510 470 480 330 72 125 230 2870 270 23 120 2370 300 680 360 24 170 55 125 75 8520 7030 15500
0.073 0.070 0.38 0.25 0.57 0.47 0.35 0.43 0.13 0.13 0.46 0.60 0.63 0.09 0.49 1.80 0.66 1.75 1.30 0.15 0.94 0.35 1.15 0.90 4.52 9.60 14.1
0.053 0.083 0.36 0.25 0.59 0.53 0.45 0.52 0.18 0.23 0.57 0.50 0.76 0.09 0.51 1S O 0.59 1.40 1.05 0.14 0.66 0.20 0.65 0.47 5.18 7.16 12.3
0.005b
0.075 0.040 0.039 0.713 0.094 0. 76 0.25 0. 18 0.25 0.077 0.28 0.39 0.43 0.063 0.20 0.2 7 0.29 0.58 0.39 0.36 0.52 0.63
0.077 0.064 0.30 0.20 0.48 0.38 0.33 0.34 0.12 0.17 0.34 0.37 0.37 0.07 0.36 1.10 0.41 1.05 0.77 0.08 0.61 0.18 0.67 0.51
0.076 0.045 0.19 0.12 0.26 0.21 0.17 0.17 0.05 0.07 0.18 0.20 0.19 0.03 0.17 0.57 0.20 0.51 0.40 0.04 0.43 0.10 0.34 0.27
3.60 5.75 9.3
1.93 3.06 5.0
0.003 0.008 0.079 0.077 0.045 0.035 0.05 7 0.069 0.056 0.078 0.007 0.070 0. 13 0. 74 0.024 0.067 0.075 0.710 0. 17 0.25 0. 78 0.27 0.36
Italicized a Equilibrium concentrations of dissolved hydrocarbons when an excess of La Rosa crude oil from the spill tank was mixed with sea water at 25 O C . value is percent hydrocarbon found in water sample compared with equilibrium concentration of dissolved hydrocarbon (oil max). Number is relative retention time (see Figure 7).
the dispersed oil droplets. This 18-min sample had the highest observed oil content (17.8 ppm) of all the subsurface samples collected. The decreasing percentage with decrease in carbon number (column 3) confirms data in Table I11 showing that these measured hydrocarbons are residual in the emulsion droplets. Note again that solution preferentially removed benzene and toluene (and probably the higher aromatic hydrocarbons, but to a lesser extent) from these droplets compared with corresponding-carbon-number saturates. Weathering of La Rosa Crude Oil. Table V, for immediately dispersed La Rosa crude oil, shows the concentrations of low-molecular-weight hydrocarbons in water samples collected a t 0 and 3 m, and 47 and 94 min after dispersion. Also shown are the equilibrium concentrations of dissolved hydrocarbons attainable when an excess of La Rosa crude oil was thoroughly mixed with sea water (column 1,oil max), and the percentage of hydrocarbons remaining in the dispersed droplets (columns 4 and 7). The equilibrium concentrations (oil max) for La Rosa crude oil are somewhat different from Murban, reflecting the differences in specific gravities and viscosities. La Rosa has less CR+ hydrocarbons than Murban, but comparable ( 2 1 4 2 4 . The lower Cg+ and aromatic hydrocarbon concentrations reflect the lower naphtha fraction (La Rosa, 11~ 0 1 %Murban, ; 19%) (2). The comparable C1-C4 concentrations are probably re-
lated to less complete separation of gas from the more viscous La Rosa during oil-field processing. The hydrocarbons in water samples also reflect the apparently slower diffusion (evaporation and solution) from the more viscous La Rosa crude oil, particularly for the CI-C~ fraction. There also was a slower change in concentration with time and depth, compared with Murban. This slower weathering may be due not only to higher viscosity but also to larger droplet sizes. Observations and extractable oil reported previously show La Rosa to have been less effectively dispersed, compared with the almost complete dispersion of Murban. However, the generally lower concentrations were more uniformly dispersed to 6 m (Table 11).The larger La Rosa droplets being less buoyant (0.91 g/mL) may well have mixed downward to this depth by wave action more easily than Murban. Although the weathering of La Rosa was slower, it should also be noted that the concentrations of low-molecular-weight hydrocarbons were very low in the water samples. The highest concentrations of an individual hydrocarbon were 1.5 ppb toluene a t 3 m, 47 min after dispersion, and 0.6 ppb a t 3 m, 94 min after dispersion. Total low-molecular-weight hydrocarbons were less than 15 ppb for all samples. The percentages of benzene and toluene (columns 4 and 7 ) show, as for Murban crude oil, that these aromatics were preferentially removed by solution from the oil droplets. Volume 14, Number 12, December 1980
1517
However, once removed, they were apparently very quickly diluted or evaporated to the atmosphere, as previously discussed. S u m m a r y a n d Conclusions
Four research oil spills of two crude oils were made off New Jersey. Two spills were immediately sprayed by helicopter with a self-mix dispersant; two, after 2 h. Average total oil determined by IR 20-60 min after dispersion under the immediately dispersed slicks a t 1,3, 6, and 9 m were 0.7,0.7,0.3, and 0.2 mg/L for La Rosa crude oil and 3.1, 2.4,0.5, and 0.4 mg/L for Murban crude oil. The highest concentrations 20-45 min after dispersion were 3 mg/L (La Rosa) and 18 mg/L (Murban). Oil concentrations for dispersion delayed 2 h were only slightly higher than found under nondispersed oil sampled immediately after discharge. The less effective dispersion after delayed treatment reflects lower and less efficient dispersant application to the thicker oil portion of these small spills, as well as increased oil viscosities due to weathering. Water samples collected 2-4 h after dispersion contained no more than two to three times background concentrations of -0.06 mg/L. Rough material-balance calculations, supported by visual and photographic evidence, indicated that Murban crude oil treated immediately was almost completely dispersed; for La Rosa, about half was dispersed. I t follows that oil removed from the influence of wind will not travel as far, and thereby will reduce the likelihood of oil stranding or entering biologically sensitive areas. The dispersed oil in the water column weathered very rapidly. Evaporation of C1-Clo hydrocarbons greatly exceeded solution. Relative concentrations of the individual C1-Clo hydrocarbons show that dissolved hydrocarbons (including benzene and toluene) did not exceed the method detection limit of 0.01 pg/L. Apparently the more soluble hydrocarbons quickly evaporated or diluted to even lower concentrations. The measured C1-Clo hydrocarbons were residual in dispersed oil droplets, and their sum did not exceed 50 pg/L, even for samples collected a t 1m and 18 min after dispersion. After 2 h they had decreased to
