Metal contents in crudes much lower than expected

Chris Dekkers
Ministry of Housing, Spatial Planning, and the Environment (VROM)
The Hague

Rinus Daane
Shell International Oil Products BV
Amsterdam

Figure 1 [137,050 bytes]
Figure 2 [151,281 bytes]
A study of crude oils imported into The Netherlands analyzed the quantity and the origins (whether indigenous or contaminants) of certain metals in the crude. This report shows that heavy metals' contents of crudes are much lower than previously assumed.

Published literature in the late 1980s and early 1990s had already identified the source and quantity of metals emissions in The Netherlands.¹² Of those metals identified, only the lead and nickel data were considered precise.

For cadmium, zinc, copper, chromium, arsenic, and mercury, the data showed a large scatter. Consequently, the uncertainty of the assessment was considered high. Suspected reasons for the large scatter are improper analytical techniques and/or contamination during sampling.

Thus, this study, started in 1991, focused on cadmium, zinc, copper, chromium, and arsenic. The intention to study mercury was abandoned because a reliable analysis technique was not found.

This study demonstrated that cadmium, zinc, and copper are not indigenous to the crude-hydrocarbon mixture. Instead, these metals are the result of contamination with associated water and/or particles from the producing wells.

Although some inorganic chromium may be present as a contaminant, chromium, for the most part is indigenous to and associated with the hydrocarbon matrix. Nickel and vanadium are similar to chromium in this way.

Although questions remain about the origins of arsenic, it probably should be considered a contaminant.

Table 1 [69,195 bytes] summarizes the properties of crudes imported into The Netherlands and the trace metals associated with these crudes.

A portion of inorganic compounds is probably removed during settling in refinery storage and desalting. On the other hand, some pick-up during processing might occur, especially for copper, zinc, and chromium.

Despite rigorous precautions to ensure representative sampling, some contamination took place during discharge and inline sampling. Hence, pick-up from sampling equipment and containers is the most probable cause for the large scatter and high numbers reported in this study as well as in previous literature.

It should be noted that pick-up in relatively small sample containers will be significantly higher, due to the larger surface to volume ratio, than in large storage tanks and process equipment.

Except for arsenic, the metals investigated are expected to concentrate in residues when passing through the desalters. Arsenic, which could be present as a volatile compound, may turn up in distillate fuels.

This project was financed by the Dutch Ministry of Housing, Spatial Planning, and the Environment (VROM), The Hague. A steering committee supervised, directed, and guided the project.

Sampling

The sampling strategy ³ was designed so that it is possible to pinpoint the probable contamination source in most contamination situations. Sampling took place at the main import terminals in The Netherlands:

From all relevant cargoes, an inline sample was taken. This yielded an average base value for the total cargo and was indicative of the amount of heavy metal imported into The Netherlands for crude oil processing and combustion.
Two randomly chosen tanks on the ship were manually sampled as back-up samples in case the inline sample had unexpectedly high values. This allowed pinpointing of a possible contamination source: either one of the ship's tanks or the discharge pipeline and related sampling system.
From the two chosen tanks, a top and bottom sample were taken. This allowed discernment between indigenous hydrocarbon-bound metals (homogeneously distributed through the tank) and inorganic contaminant metals, which are present as particles or dissolved in the associated water (inhomogeneously dispersed through the tank).
At least one cargo of each Dutch-processed crude was sampled. For the two most prevalent crudes processed in Dutch refineries (Table 2 [45,592 bytes]), five cargoes were sampled over 1 year. This allowed determination of cargo-to-cargo variability and assessment of the effect of seasonal influences (for example, field maintenance and well workovers) on the metal levels.

A bottom-sampling system was designed and manufactured out of glass and titanium to exclude avoidable contamination sources. ⁴

Samples were collected in acid-washed, clear glass bottles with Teflon-inlaid screw caps. The samples were stored under controlled (cold and dark) conditions until needed.

Table 3 [105,063 bytes] lists the type of samples taken and the basic properties of the crudes sampled from December 1993 to February 1996.

Analysis

Table 4 [83,277 bytes] lists the applied analytical techniques which were chosen based on the expected concentration and type of material to be analyzed. All methods were optimized and validated before actual samples were analyzed. ⁵

The analytical work was contracted to an independent institute, the Dutch Organization for Applied Natural Scientific Research (TNO, Delft, The Netherlands). The sampling was subcontracted to a qualified cargo inspector, Saybolt Nederland, Vlaardingen, The Netherlands.

Each cadmium sample was analyzed ten times; the other metal samples were analyzed five times. Averaged results for the inline crude samples are reported in Table 5 [53,699 bytes].

Cadmium

The cadmium contents of sampled crudes were very low (Table 5). Average cadmium contents ranged from <0.4 ?g/kg (detection limit) to 2.4 ?g/kg.

A frequency plot of the measured cadmium data showed a log-normal distribution. This suggests that cadmium is a contaminant rather than an organic-bound element in the hydrocarbon matrix.

This conclusion is further substantiated by Table 6 [36,223 bytes], which shows that the range of cadmium content for Arabian Light crude over five different cargoes is about equivalent to (actually slightly larger than) the range for all crudes.

Furthermore, the analyses of the top, bottom, and inline sample indicate that cadmium is not homogeneously distributed through the tanks. Rather, its concentration increases from top to bottom, which indicates that cadmium is either present as (inorganic) particles or dissolved in the water phase.

Linear regression of the averaged cadmium data showed that the cadmium data are best represented by an inverted, linear function of the log sulfur content of the crude (r2=0.95; Fig. 1a).

The present work is in good agreement with the other sources, except that by Al-Swaidan (Table 7 [85,372 bytes]).

Zinc

Average values of zinc from this study range from 116 ?g/kg to 500 ?g/kg (Table 5).

Like cadmium, the zinc data fit a log-normal distribution (Fig. 1b); hence, it is probably also a contaminant.

This conclusion is further substantiated by Table 6, which shows that range of zinc values for Arabian Light crude over five different cargoes is about equivalent to (actually slightly larger than) the average range for all crudes.

Furthermore, analyses of the top and bottom samples indicate that zinc is not homogeneously distributed through the tanks. The concentration increases from top to bottom, which indicates that zinc is either present as (inorganic) particles or dissolved in the water phase.

Unlike cadmium, however, the zinc content of the inline sample is actually higher than the zinc content of the bottom sample. Probably, further contamination with zinc took place between the ship's railing and the moment the samples for metals analysis were taken.

Some evidence for this conclusion is presented in Table 8 [53,969 bytes], which shows the zinc contents of inline samples taken from the same cargo but at different terminals (multiple discharge cargoes). For five of seven cases, the zinc content of the inline samples taken at the Texaco-Esso Europoort terminal (TEM) are significantly higher then those taken at the Maasvlakte oil terminal (MOT).

Linear regression of the averaged zinc showed that the zinc data are best represented by a linear function of the sulfur content of the crude (r2=0.92; Fig. 1b). Only Arabian Light appears as an outlier.

Zinc and cadmium are consistently found together in nature in Zn/Cd ratios between 40 and 800.⁶ With the exception of Arabian Heavy and Kuwait (both 800), all Zn/Cd ratios fall between those limits.

Regression of the Zn/Cd ratios with the available crude oil properties shows that the ratio can be predicted from the sulfur content of the crude (r2=0.93; Fig. 1c) with the exception of Arabian Light crude.

Interestingly, a plot of the ratios of the inline samples for the five cargoes of Arabian Light over time shows a cyclic, seasonal behavior (Fig. 1d). The ratio reaches its theoretical value about mid-winter.

This peak could point to artifacts introduced by oil field operations. Well stimulation by acids temporarily changes the pH of the associated water and consequently affects the relative solubility of zinc and cadmium compounds.

The zinc data are positioned at the lower end of the scale when compared to other data (Table 7). It is likely that a significant amount of the high (1,000 ?g/kg) data are caused by contamination of zinc from sampling equipment and cans.⁷⁸

Copper

The data for copper, although lower, resemble that for zinc. The data fit a log-normal distribution. The range covered by the Arabian Light cargoes is approximately the same as that for all individual crudes. Moreover, a significant difference exists between top and bottom samples.

Consequently, copper is likely to be present as an inorganic contaminant in crude. According to this study, average values for copper content range from 21 ?g/kg to 195 ?g/kg.

Like zinc, the copper content of the inline sample was higher than the copper content of the bottom sample. As mentioned before, further contamination with copper probably took place between the ship's railing and the moment the samples for the metal analysis were taken.

Also similar to zinc, the inline samples taken from the same cargo but at different terminals show that, in general, samples taken at TEM have significantly higher copper contents then those taken at MOT.

Linear regression of the averaged copper data showed that the copper data are best represented by a second degree function of the sulfur content of the crude (r2=0.85; Fig. 2a). The relationship, however, is not as strong as for cadmium and zinc. Moreover, there are two outliers, Ural and Oseberg.

Remeasurement of some samples confirmed the previous analytical results (Table 9 [34,090 bytes]).

The Statfjord crude data confirmed that the terminal does play a role. On average, the TEM terminal yielded results that were twice as high. The Oseberg remains an outlier, which cannot be explained from differences in pick-up through sampling equipment at the terminals alone although the Oseberg sample was the only one sampled at the Shell terminal. The relatively high values for the bottom sample indicate that contamination must have taken place before or during loading of the cargo.

A comparison of the copper data with the previously published literature data is shown in Table 7. The present data are the lowest in the listing. A plausible explanation of the relatively high copper data in literature is that, as shown, copper is easily picked up from various sources. For example, the earth's crust contains an average of 125 mg/kg copper.⁹

Furthermore, traditional oil-sampling equipment is made of copper or brass, and many sample cans contain copper.

Finally, it is noteworthy to observe that Oseberg has the highest (organic) acidity (total acid number=0.15 mg KOH/g) of all the crude oils analyzed. In previously published data, crudes with high acidities also have the highest copper contents, which suggests corrosion may be a pick-up mechanism.

Chromium

In comparison to the three foregoing metals, the chromium data show a distinctly different frequency distribution. The chromium distribution is clearly bimodal. This indicates that, unlike the three foregoing metals, chromium is present, for the main part, as an organometallic compound like the nickel and vanadium complexes.

Furthermore, the range covered by the individual Arabian Light cargoes (19-27 ?g/kg) is significantly narrower then the range covered by the individual crude oils (<13-161 ?g/kg).

On the other hand, a comparison of the analysis of the top/bottom samples with the inline samples and the existence of a few high results, related to the discharge terminal, indicate that both indigenous chromium and contaminant chromium exist (Table 10 [41,387 bytes]). Pick-up of chromium, like zinc and copper, most probably takes place between the ship's railing and the moment the samples for metal analyses are taken.

Linear regression of the averaged chromium data shows that the chromium data are best represented by an exponential function of the vanadium content of the crude (r2=0.99; Fig. 2b). No unexplainable outliers exist.

A comparison of the chromium data with the previous literature data shows the present data are among the lowest (Table 7). Again, the most likely cause of some of the very high data reported in previous literature is contamination (pick-up) through sampling equipment and containers.

Arsenic

In contrast to the other metals, the picture for arsenic (As) is less clear cut. Analytical results were nondiscriminating. Except for one relative high value (37 mg/kg), all other data range from 16 ?g/kg to <10 ?g/kg. The statistical detection limit is 10 ?g/kg.

The raw data are best represented by a log-normal distribution. Unlike the other contaminant metals, however, the scatter per individual crude is less then over the whole set. Consequently, it is difficult to draw a hard conclusion regarding its speciation (organic or inorganic bound).

For two cargoes (Arabian Light and Ural) the top and bottom samples of the ship's tanks were analyzed. This analysis revealed no clear trends: The three Arabian Light samples (inline, top, and bottom) all yielded a value of <10 ?g/kg As, and the Ural cargo consistently yielded a value around 37 ?g/kg.

Linear regression of the averaged arsenic data (with estimates for the data below the statistical detection limit) showed that for the available crude oil properties, the best fit was obtained with the crude oil viscosity (log). Two outliers remain, however: Ural (too high) and Arabian Heavy (too low). A repeat of the analyses of these cargoes confirmed the results.

A second sample of Arabian Heavy showed an arsenic content of 23 ?g/kg (ten five-fold determinations over the course of the project), and this value fits exactly on the regression line (r2=0.93; Fig. 2c). This suggests that arsenic is present as an inorganic contaminant and part of the bs&w (bottoms, sediment, and water). Reliable bs&w data, however, were not available to check whether indeed a correlation with bs&w exists.

In Table 11 [43,091 bytes], the relative abundance of arsenic vs. the other contaminants is tabulated with the expected ratio based on the relative abundance of the elements in the earth crust.¹⁰

Project data showed a good correlation between arsenic and zinc (r2=0.86; Fig. 2d). The average arsenic/zinc ratio was an almost perfect match with the relative natural abundance of these metals. Moreover, the coefficient of variance is very low, indicating that the ratio is constant for all crudes tested (except for Ural) and not correlated with any of the crude oil properties.

The only other metal ratio for which a match is obtained with the natural abundance is the zinc/cadmium ratio. However, here the co-variance is significantly higher, and it was shown earlier that this ratio has a strong correlation with the sulfur content of the crude.

Most probably, arsenic is present as an inorganic contaminant and strongly related, to and probably in association with, zinc. The arsenic/zinc ratio for the Ural cargo (0.10) differs significantly. This, in combination with the fact that for this crude the concentration for top and bottom samples proved to be same, would imply arsenic is probably present as an oil-soluble compound in Ural crude.

Without further work, it is not possible to conclude whether the arsenic present in Ural is indigenous to the crude or a man-made contaminant.

Table 7 shows that the present arsenic data are among the lowest. In the previously published literature data, the high values are associated with high viscosity crudes and/or crudes that are known to contain significant amounts of bs&w. High viscosity crudes are usually difficult to dehydrate and desalt.

Acknowledgments

The authors thank the OCC (Oil Contact Committee) for its permission to initiate the project, the oil companies for providing the samples, the steering committee for guiding this project, and VROM for its permission to publish this work. Furthermore, we thank the contractors and subcontractors for the thorough sampling and analyses.

References

Stobbelaar, G., "Reduction of Atmospheric Emissions under the terms of the North Sea Action Programme," June 1992.
Veldt, C., "Emission factors for micropollutants from fired processes," February 1990.
Spijk, S.J., et al., "Determination of the Cadmium Content of Various Types of Crude Oil," TNO-MEP-R 96/189, TNO Institute of Environmental Sciences, Energy Research and Process Innovation, Sept. 16, 1996, Apeldoorn, The Netherlands.
Spijk, S.J., "Protocol Crude Oil Sampling," TNO-IMW-R 93/211, TNO Institute of Environmental Sciences, Department of Analytical Chemistry, Aug. 31, 1993, Delft, The Netherlands.
Submitted for publication in Environmental Pollution.
Schroeder, H.A., "Air Quality Monographs, Monograph No 70-16, Cadmium, Zinc and Mercury," American Petroleum Institute.
Ang, C.C., Zhuguo, D., Jiamo, F., and Guoying, S., "Determination of Trace Elements in Crude Oils and Organic Materials Extracted from Oil-forming Source Rocks in China by INAA," J. Radioanal. Nuclear Chem., Articles, Vol. 151, No. 1, 1991, pp. 177-84.
Jones, P., "Trace Metals and Other Elements in Crude Oils, A Literature Study," BP Research Centre Sunbury, U.K., September 1975.
Goldschmidt, V.M., Geochemistry, Clarendon Press, Oxford, 1954.
Mason, Brian, Principles of Geochemistry, 1952.
Stigter, J.B., et al., "Determination of the Zinc, Arsenic, Copper and Chromium Content of Various Types of Crude Oil," TNO-MEP-R 97/095, TNO Institute of Environmental Sciences, Energy Research and Process Innovation, Nov. 3, 1997, Apeldoorn, The Netherlands.
Narrres, H.D., Mohl, C., and Stoepler, M., "Metal Analysis in Difficult Materials with Platform Furnace Zeeman-Atomic Absorption Spectrometry," Intern. J. Environ, Anal. Chem., Vol. 18, 1984, pp. 267-79.
Al-Swaidan, H.M., "Simultaneous Determination of Trace Metals in Saudi Arabian Crude Oil Products by Inductively Coupled Plasma Mass Spectrometry (ICP/MS)," Analytical Letters, Vol. 21, No. 8, 1988, pp. 1487-97.
Valcovic, V., Trace Elements in Petroleum, Tulsa, 1978.
Vliet, H.P.M. van, and Feenstra, J.F., IVM-VU, Amsterdam, 1982.
Anwar, J., Nahid, F., and Kirmani, Z.U., "Determination of Trace Metals in Crude Oil Samples by Atomic Absorption Spectroscopy Using a Mixed Solvent System," J. Chem. Soc. of Pak., Vol. 12, No. 3, 1990, pp. 249-52.
Colombo, U.P., Sironi, G., Fasolo, G.B., and Malvano, R., "Systematic Neutron Activation Technique for the Determination of Trace Metals in Petroleum," Anal. Chem., Vol. 36, No. 4, 1964, pp. 802-07.
Osibanjo, U., Kakulu, and S.E., Ajayi, S.O., "Analytical Application of Inorganic Salt Standards and Mixed-solvent Systems to Trace-metal Determination in Crudes by Atomic-absorption Spectrophotometry," Analyst, Vol. 109, 1984, pp. 127-29.
Hitchon, B., Filby, R.H., and Shah, K.R., "Geochemistry of Trace Elements in Crude Oils, Alberta, Canada," American Chemical Society, Div. Petr. Chem., Inc., Chicago meeting, Aug. 26-31, 1973, Vol. 18, No. 4, pp. 623-29.
Shah, K.R., Filby, R.H., and Haller, W.A., "Determination of Trace Elements in Petroleum by Neutron Activation Analysis," J. Radioanal. Chem., Vol. 6, 1970, pp. 413-22.
Filby, R.H., and Shah, K.R., "Determination of Cadmium in Petroleum by Neutron Activation Analysis," American Chemical Society, Div. Petr. Chem., Inc., Chicago meeting, Aug. 26-31, 1973, Vol. 18, No.4, pp.
615-18.
Veal, D.J., "Nondestructive Activation Analysis of Crude Oils for Arsenic to One Part per Billion, and Simultaneous Determination of Five Other Trace Elements," Anal. Chem., Vol. 38, No. 8, 1966, pp. 1080-183.
Ali, M.F., Bukhari, A.B., and Saleem, M., "Trace Metals in Crude Oils from Saudi Arabia," Ind. Eng. Chem. Prod. Res. Dev., Vol. 22, No. 4, 1983, pp. 691-94.
Olajire, A.A., and Oderinde, R.A., "Trace Metals in Nigerian Crude Oils and Their Heavy-End Distillates," Bull. Chem. Soc. Jpn., Vol. 66, No. 2, 1993, pp. 630-32.
Araktingi, Y.E., Chakrabarti, C.L., and Maines, I.S., "Determination of Trace Elements in Crude Oils by Nonflame Atomic Absorption Spectroscopy," Spectroscopy Letters, Vol. 7, No. 2, 1974, pp. 97-120.
Gonz lez, M.C., Rodriguez, A.R., and Gonz lez, V., "Determination of Vanadium, Iron, Copper, and Lead in Petroleum Fractions by Atomic Absorption Spectrophotometry with a Graphite Furnace," Microchem. J., Vol. 35, 1987, pp. 94-106.
Speight, J.G., The Chemistry And Technology Of Petroleum, Second Edition, ISBN 0-8247-8481-2, Marcel Dekker Inc., New York, 1991.
Filby, R.H., and Olsen, S.D., "A Comparison of Instrumental Neutron Activation Analysis and Inductively Coupled Plasma-Mass Spectrometry for Trace Element Determination in Petroleum Geochemistry," J. Radioanal. Nuclear Chem., Vol. 180, No. 2, 1994, pp. 285-94.
Jackson, K.F., Benedik, J.E., and Birkholz, F.A., "Metals Distribution in Shale Oil Fractions," Oil Shale Symp. Proc., Vol. 14, 1981, pp. 75-81.
Walker, H.H., Runnels, H.H., and Merryfield, R., "Determination of Trace Quantities of Selenium in Petroleum Products by Atomic Absorption Spectrometry," Anal. Chem., Vol. 48, No. 14, 1976, pp. 2056-60.
Bergerioux, C., and Zikovsky, J.J., "Determination of 18 Trace Elements in Petroleum and its Derivatives by Neutron Activation with a Small Nuclear Reactor," Radioanal. Chem., Vol. 46, 1978, pp. 277-84.
K?gler, S. et al., "Determination of Cadmium in Petroleum Products," Deutsche Gesellschaft f?r Mineral?lwissenschaft und Kohlechemie E.V., DGMK-Project 264, August 1983, Hamburg, Germany.
Robbins, W.K., and Walker, H.H., "Analysis of Petroleum for Trace Metals-The Determination of Trace Quantities of Cadmium in Petroleum by Atomic Absorption Spectrometry," Anal. Chem. Vol. 47, No. 8, 1975, pp. 1269-75.
Hofstader, R.A., Milner, O.I., and Runnels, J.H. (eds.), Analysis of Petroleum for Trace Metals, Advances in Chemical Series 156, American Chemical Society, Washington D.C., 1976.

The Authors

Chris Dekkers is senior policy advisor in the Air & Energy Directorate of the Ministry of Environment, The Netherlands. He is responsible for the development of strategy and policy decisions on industrial air emission reductions, in particular that of large combustion plants. He also coordinates environmental policy decisions for the refining industry, and participates in the National Emission Task Force for the European Auto-Oil Programme. Dekkers has previously worked for the Oil Directorate of Ministry of Economic Affairs, where he was involved in refining projects and standards, and the United Nations, where he worked on development programs for Cameroon and Tanzania. Dekkers holds a degree in mathematical economics from the Erasmus University of Rotterdam.

Rinus Daane is senior research chemist at Shell Research & Technology Centre in Amsterdam. As project manager of the crude oil and product department, he develops analytical methods on crude oil and oil products and provides technical support on product/feedstock characterization and processability. Daane is an active member of various organizations, including American Society for Testing and Materials (ASTM D2), Institute of Petroleum (IP), International Standardization Organization/Petroleum Products (ISO/TC28), (Commite Europeen de Normalisation (CEN/TC19), and the Dutch Standardization Institute (NNI).