Concentration of Mercury in Crude Oil Refined in the United States
Wilhelm, S. M., and Bigham, G. N., 5th International Conference on Mercury as a Global Pollutant,
Minamata, Japan, 2002.
Introduction
We have compiled and evaluated data on
the concentration of mercury in crude oil processed by U.S. refineries. The range of total mercury concentrations in crude oil spans 5 or 6 orders of magnitude, from parts per trillion (ppt) to parts per million (ppm) concentrations. The exact range of total mercury concentrations in crude oil cannot be assigned at present, and a defensible mean concentration can only be stated very approximately because of uncertainty in the quality of the analyses and the representativeness of the results.
Several recent studies that have examined the high volume sources of crude oils processed in the United States and Canada suggest that the mean amount in U.S. processed crude is much less than previously estimated by the U.S. Environmental Protection Agency (EPA) (U.S. EPA 1997a,b). Currently available data, corrected by discounting two anomalous fields and without weighting for volume, suggest the mean should be less than 20 parts per billion (ppb).
In virtually all of the studies of mercury in crude oils that were examined, samples were poorly documented as to their source and handling before analysis, leaving the data derived from them in doubt. As has been found for mercury in other matrices, sampling method and post-retrieval processing can lead to order of magnitude variations in measured mercury concentrations. In addition, analytical uncertainties exist in some data because of unreported or incomplete quality assurance procedures. In general, however, analytical methods, as applied to mercury in hydrocarbon matrices, generally have improved in the last decade by incorporating aspects of the newer procedures developed for measuring (and speciating) mercury in water.
A few crude oils in the world have high (ppm) concentrations of mercury due to the peculiarities of their geologic origin. Because of the difficulties they present in processing and recent environmental concerns, analyses of such high-mercury crude oils abound, and the frequency of their inclusion in studies is disproportionate with the volumes they represent in U.S. refining.
Mercury in Crude Oil
Crude oil typically contains several chemical forms of mercury, which differ in their chemical and physical properties (Wilhelm and Bloom 2000; Bloom 2000). Dissolved elemental mercury (Hg0) is soluble in crude oil and hydrocarbon liquids in atomic form to approximately 2 ppm. Dissolved organic mercury compounds (RHgR and RHgY, where R = CH3, C2H5, etc., and Y = Cl-, etc.) are highly soluble in crude oil and gas condensate. This category includes dialkylmercury (i.e., dimethylmercury, diethylmercury) and monomethylmercury halides (or other inorganic ions). Dialkylmercury compounds have been detected in condensates, but their natural abundance across classes of hydrocarbon liquids is unknown. Monoalkylmercury is seldom detected in oil and condensate, inferring that the dialkyls are likewise not abundant.
Other dissolved forms of mercury in crude oil include inorganic (ionic) mercury salts (Hg2+Z or Hg2+Z2, where Z is an inorganic ion) that are soluble in oil and gas condensate but preferentially partition to the water phase in primary separations. Mercuric chlorides have a reasonably high solubility in organic liquids (about 10 times more than elemental mercury). Complexed mercury (HgK or HgK2) can exist in hydrocarbons as a complex, where K is a ligand (probably containing sulfur) such as a thiol or thiophene. The existence of such compounds is inferred from operational analytical protocols, but their exact identity and prevalence are a matter of speculation at present. In addition to the purely dissolved forms, crude oil typically contains suspended mercury compounds or mercury adsorbed on suspended solids or both. The analysis for total mercury in crude oil yields the sum of both dissolved and suspended forms. Although the concentration of the truly dissolved forms of mercury is thought to be reasonably consistent within a given reservoir, the concentration of suspended forms depends on the location from which samples were taken in the production, transportation, or refining process. The percentage of suspended mercury in the measured total amount in crude oil typically decreases with distance from the wellhead because of its loss in separators, pipelines, and tanks.
In most crude oil production, three phases separate (water, gas, and oil) in equipment immediately downstream of the wellhead. The amount of mercury in the oil phase depends in part on the relative amounts of the other phases, which change over time as the reservoir depletes. Elemental mercury typically equilibrates between gas and liquid hydrocarbon and other forms (oxidized, complexed, organic) remaining mostly in the liquids. The partitioning of mercury in primary separations is largely determined by solubilities and vapor pressures of the various species.
Historical Data
The accuracies of concentrations reported for mercury in crude oils are difficult to assess because of the lack of documentation of sample source and handling before analysis. In general, sampling errors lead to lower than actual values of THg in petroleum. The main reasons for lower than actual concentrations include the following:
- Loss of volatile Hg0 during sampling to the headspace of sample containers
- Failure to account for mercury adsorbed to sample containers
- Failure to obtain homogeneous samples that are representative of the actual suspended fraction.
If the true concentration of mercury in a crude oil is very low (<10 ppb), analytical variations can be positive by inclusion of (laboratory) atmospheric mercury and addition of mercury in impure reagents. Analytical methods sometimes produce negative bias from loss of volatile mercury in hot digestions, incomplete digestions, or inefficient extractions. For these reasons, data quality is difficult to assess quantitatively.
The range of concentrations of mercury in crude oil appears to span ppt to ppm levels. The lower end of the range is uncertain because the detection limit for mercury in crude oil achieves sub-ppb levels only for cold vapor atomic fluorescence (CVAF) detection methods, but data obtained by using these methods do not always agree. The method detection limit (MDL) for neutron activation analysis (NAA) remains close to 1 ppb but probably could be improved with method development efforts directed specifically at mercury. NAA is still considered the standard analytical method to which other methods are compared, but no comparisons of ability to measure mercury in hydrocarbon matrices have been conducted for the newer methods (combustion CVAF, extraction CVAF, inductively coupled plasma-mass spectrometry) relative to NAA.
Statistical Considerations
A major impediment to assigning a mean concentration to mercury in crude oils processed in the United States is that representative concentrations must be weighted for the volumes that are processed. At present, it is not certain that concentrations reported for particular fields are representative. Although production statistics for U.S. fields exist, the percentages of foreign field production processed in U.S. refineries are less well known. Such statistics might be developed, but they are not readily available now. Additionally, it is not known if mercury concentrations assigned to a field are representative of that field, given the mixing that occurs in multiple well completions and in gathering lines. Furthermore, characterization of mercury in foreign oil carried by tanker may not be representative of the source fields because of mixing from different sources at the loading terminal.
If mercury concentrations were assigned on a regional basis, gross errors would be expected unless sufficient data existed to allow a statistical interpretation. This expectation is clearly shown in the work of Hitchon and Filby (1983), who examined mercury in Alberta (Canada) crudes across the range of regional geologies. The range for the 86 crudes examined spanned amounts less than detection limits (2 ppb) to approximately 400 ppb, and close to half the reported values were less than the detection limit. Thus, assignment of a mean concentration to a country or region would require a statistical study such as Hitchon's that examined local geologies and production volumes.
Discussion
Calculated mean concentrations of mercury in crude oil processed domestically that are applied by U.S. regulatory bodies are in error. The principal reason is that much of the existing data on mercury in U.S. crude oils contain an inordinate contribution of measured concentrations from a small field in California. Mercury concentrations from this field abound in the earlier data because of the interest of analysts. This single field contributes approximately 17 of the 5,000 million barrels of crude oil processed annually in the United States (foreign and domestic) or about 0.34 percent (U.S. DOE 2000 for year 1999). Reported concentrations of mercury from this field, however, account for more than 10 percent of the data used by EPA to calculate an average concentration of mercury in crude processed in the United States. Given that crude from this small California field has the highest reported concentration of mercury in the world, current estimates are at least 30 times higher than they would be if calculations of an average were based on volume-weighted contributions.
Inclusion of Bloom's data (Bloom 2000) when calculating a mean concentration for mercury crude oil creates a similar bias. Such calculations, if based on number of samples analyzed and not on normalized process volumes in a particular ensemble, lead to gross errors in calculating the mean amount of mercury in crudes processed in the United States. The upper half of Bloom's data on mercury in crude oil come from one field producing less than 10 million barrels per year or about 0.2 percent of crude processed in U.S. refineries annually (assuming this crude is processed in the United States, which is not actually known). This one field accounts for more than 30 (of about 300) data points in calculated averages.
These points are illustrated in Table 1. By discarding data from two fields that account for at most 0.6 percent of the crude processed in the United States, one obtains an adjusted mean close to 20 ppb instead of 600 ppb. The adjusted mean of 20 ppb is probably higher than the actual mean concentration of mercury in crude oil because it includes the early NAA data whose sample sets were selected for analytical method development.
Table 1. Summary of mercury concentrations in crude oils
| Method |
Detection Limit
(ppb) |
Reference |
Number of Samples |
Low
(ppb) |
High
(ppb) |
Mean
(ppb) |
Adjusted
Mean
(ppb) |
| TD-CVAF |
0.2 |
Liang (2000) |
9 |
1.0 |
9.3 |
3.1 |
3.1 |
| TD-CVAF |
|
|
11 |
1.6 |
7.2 |
3.4 |
3.4 |
| E-CVAF |
0.5 |
Bloom (2000) |
76/37 |
- |
- |
1,500 |
1 |
| E/TD-CVAF |
0.5/0.2 |
Morris (2000) |
23 |
0.1 |
12.2 |
3.2 |
3.2 |
| ICP-AES |
15 |
Cao (1993) |
24 |
<15 |
<15 |
8 |
8 |
| D-CVAA |
10 |
Magaw et al. (1999) |
26/25 |
<10 |
1,560 |
65 |
5 |
| D-CVAA |
2 |
Duo (2000) |
8 |
<2 |
9 |
1.6 |
1.6 |
| D-CVAA |
10 |
Knauer and Milliman (1975) |
3 |
<10 |
<10 |
5 |
5 |
| NAA |
4 |
Shah et al. (1970) |
10/9a |
23 |
29,700 |
3,200 |
297 |
| NAA |
2 |
Hitchon et al. (1983) |
86 |
<2 |
399 |
21.9 |
22 |
| NAA |
2 |
Filby et al. (1975) |
4/3 |
<4 |
23,100 |
5,803 |
38 |
| NAA |
0.1 |
Musa et al. (1995) |
6 |
0.1 |
12.2 |
3.1 |
3.1 |
| TOTAL |
|
|
286/244 |
0.1 |
30,000 |
605 |
21 |
References
- Bloom, N.S. 2000. Analysis and stability of mercury speciation in petroleum hydrocarbons. Fresenius' J. Anal. Chem. 366(5):438.
- Cao, J.R. 1992. Microwave digestion of crude oils and oil products for the determination of trace metals and sulphur by inductively-coupled plasma atomic emission spectroscopy. Environment Canada Manuscript Report Number EE-140, Ottawa, Ontario.
- Duo, W., T. Guthrie, and W.C. Edwards. 2000. Mercury emissions from the petroleum refining sector in Canada. Contract: K2216-9-0078. Prepared for Environment Canada, Trans-boundary Air Issues Branch, Hazardous Air Pollutants Program, Ottawa, Ontario.
- Filby, R. H., and K.R. Shah. 1975. Neutron activation methods for trace metals in crude oil. In: The Role of Trace Metals in Petroleum. T.F. Yen (ed). Ann Arbor Science Publishers, Ann Arbor, MI.
- Hitchon, B., and R. Filby. 1983. Geochemical studies-Trace elements in Alberta crude oils. Open File Report 1983-02. Alberta Research Council for Alberta Energy and Utilities Board, and Alberta Geological Survey. Edmonton, Alberta.
- Liang, L., S. Lazoff, M. Horvat, E. Swain, and J. Gilkeson. 2000. Determination of mercury in crude oil by in situ thermal decomposition using a simple lab built system. Fresenius' J. Anal. Chem. 367:8.
- Magaw, R., S. McMillen, W. Gala, J. Trefry, R. Trocine. 1999. Risk evaluation of metals in crude oils. In: Proc. for SPE/EPA Exploration & Production Environmental Conference, SPE Paper No. 52725.
- Morris, R. 2000. New TRI reporting rules on mercury. In: Proc. of the National Petroleum Refiners Association Meeting, San Antonio, Texas.
- Musa, M., W. Markus, A. Elghondi, R. Etwir, A. Hannan, and E. Arafa. 1995. Neutron activation analysis of major and trace elements in crude petroleum. J. Radioanal. Nucl. Chem. 198(1):17.
- Shah, K.R., R.H. Filby, and W.A. Haller. 1970. Determination of trace elements in petroleum by neutron activation analysis. J. Radioanal. Chem. 6:413.
- U.S. DOE. 2000. Energy statistics for 1998, 1999. U.S. Department of Energy, Energy Information Administration, National Energy Information Center, Washington, DC.
- U.S. EPA. 1997a. Locating and estimating air emissions from sources of mercury and mercury compounds. EPA/454/R-97/012. U.S. Environmental Protection Agency, Office of Air Quality, Environmental Protection Planning and Standards Agency, Research Triangle Park, NC.
- U.S. EPA. 1997b. Mercury Study Report to Congress. EPA/452/R-97/003. U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards, Office of Research and Development, Washington, DC.
- Wilhelm, S. 2001. In press. An estimate of mercury emissions to the atmosphere from petroleum. Environ. Sci. Tech.
- Wilhelm, S., and N. Bloom. 2000. Mercury in petroleum. Fuel Proc. Technol. 63:1.
