Volume 1, Paper 12


Laboratory Study of Corrosion Effect of Dimethyl-Mercury on Natural Gas Processing Equipment

S. Wongkasemjit* and A. Wasantakorn
The Petroleum and Petrochemical College, Chulalongkorn University, Bangkok 10330, Thailand.
e-mail address :

Abstract

Dimethylmercury (DMM) has been discovered to be one of the organomercury compounds in natural gas, besides mercury metal. The effect of the DMM on the corrosion of carbon steel and aluminium metal was rather frightening, as compared to that of elemental mercury or mercuric chloride, which is well-known as a severely corrosive agent. The results showed that DMM in methanol or petroleum ether as solvent gave similar corrosion patterns to the elemental Hg solution. It revealed uniform corrosion characteristics on carbon steel and pitting appearance on aluminium. The higher the DMM concentration or the higher reaction temperature, the more severe corrosion occurred because of the higher elemental Hg concentration generated. Trace amounts of hydrogen chloride or hydrogen sulfide, which are also present in large quantities in natural gas, remarkably increased the corrosion potential of DMM on metal. Corrosion rate in the HCl+DMM solution was approximately 700 times faster than the one containing only DMM and 40 times faster than the one containing only acid.

Introduction

It is well known that ppm amounts of elemental mercury in natural gas or condensates result in corrosion of equipment in oil or natural gas processing plants1-4. This is the main cause of equipment failure, especially failure of aluminium heat exchangers. The gas separation parts of the Petroleum Authority of Thailand (PTT) at Map Ta Put has already encountered this problem, and had to be shut down for a period of 4-5 months resulting in lost income of over 10 million US dollars.

The effects of elemental mercury in natural gas on corrosion was not reported until 1973 when a catastrophic failure of aluminium heat exchangers occurred at the Skikda liquefied natural gas plant in Algeria5. It was found that mercury corrosion caused the failure. After this discovery, a study of the Groningen field in Holland revealed similar corrosion in the gas-gathering system6-7.

Organic mercury compounds are typical mercury derivatives formed along with elemental mercury in natural gas8. Most of them are liquids which are extremely volatile9, and can not be trapped by mercury absorbants. To remove them, it is essential to decompose and convert organomercury to elemental mercury either by heat and/or catalyst10-11. From the thermochemical data collected12 for mercuric chloride and dimethylmercury, mercuric chloride, which is a severe corrosive agent to metals, needs approximately 105 kcal/mol to break the bonds between mercury and chlorine atoms while the total energy required to break the bonds between methyl groups and mercury atom is approximately 58 kcal/mole to give elemental mercury. Therefore, dimethylmercury is believed to be a stronger corrosive agent.

Decomposition of dimethylmercury can be only done by either pyrolyzing or photolyzing13. One example was to decompose dimethylmercury in the presence of oxygen to give elemental mercury. It was also found that dialkylmercury could be decomposed by a mineral acid like hydrochloric acid to give mercuric chloride14, and as a result might be able to corrode natural gas equipment.

To address the question of whether dimethylmercury will corrode natural gas equipment, this research was thus focused on investigating corrosion of metals by organomercury compounds. Dimethylmercury, which is the most stable organomercury compound, was selected to study its corrosive behavior toward carbon steel and aluminium specimens.

Materials

Chemicals : All chemicals were reagent grade and used as received. Dimethylmercury (Hg(CH3)2, DMM), mercuric chloride (HgCl2) and elemental mercury (Hg) were purchased from Fluka Ag., Puriss. Petroleum ether (PE, bp. 80-100 °C), cyclohexane, 1,4 dioxane, dimethylsulfoxide, benzene, and sodium sulfide (Na2S) were purchased from E. Merck (Germany). Absolute methanol (CH3OH), antimony trioxide, and diethyl ether were obtained from J.T.Baker. Chromic acid, trichloroethylene, and isopropyl alcohol were from AJAX Chemicals Co. Phosphoric (H3PO4), hydrochloric (HCl)/sulfuric (H2SO4) acids, and stannous chloride were obtained from Carlo Erba.

Specimen Samples : Specimens, carbon steel and aluminium, were obtained from commercial carbon steel reinforcement bars and commercial aluminium bars for architectural use. They were prepared and cleaned prior to use. The preparation and cleaning was according to ASTM-G1-72.

Measurements : The compositions of both carbon steel and aluminium were analyzed by a Shimadzu QC-6 Vacuum Emission Spectrometer (Table 1). Corroded surfaces were inspected by a Scanning Electron Microscope (JEOL, JSM-35 CF).

Table 1 Compositions of commercial aluminium and carbon steel specimens tested by Vacuum Emission Spectrometer

Element

Commercial aluminium specimen (%)

Commercial carbon steel specimen (%)

Aluminium (Al)
Carbon (C)
Chromium (Cr)
Copper (Cu)
Iron (Fe)
Lead (Pb)
Magnesium (Mg)
Manganese (Mn)
Molybdinum (Mo)
Nickel (Ni)
Phosphorus (P)
Silicon (Si)
Sulphur (S)
Tin (Sn)
Titanium (Ti)
Vanadium (V)
Zinc (Zn)

98.231 
-
0.016
0.118
0.581
0.010
0.523
0.057
-
0.010
-
0.361
-
-
0.015
-
0.078
0.001
0.208
0.102
0.321
97.757
0.006
-
1.275
0.003
0.085
0.001
0.203
0.020
0.015
0.003
-
-

Methods

1. Specimen Preparation

Commercial carbon steel/aluminium bars were cut to make short strips with dimensions of 11.2 x 25 x 2.75 mm for carbon steel and 12.3 x 25 x 3 mm for aluminium. A hole was then drilled in the strips near one end for mounting. The diameter of the hole was 3.5 mm for carbon steel and 5.0 mm for aluminium. A few of these strip coupons were chosen for elemental composition analysis by vacuum emission spectrometer.

Specimens were wetted and rubbed with a No. 120 abrasive paper until their surface was cleaned and smooth. The sizes of carbon steel and aluminium specimens became 11.1 x 24 x 2.75 mm and 12.0 x 24 x 2.9 mm, respectively. All of the strip coupons were subsequently stamped with numbers on the upper left hand side near the hole. Coupons were then degreased by scrubbing with a bleach-free scouring powder, followed by rinsing with distilled water and a mixture of 1:1 ratio of methanol : diethyl ether, and finally dried with air. They were then weighed with an accuracy of 0.0001 g, and kept in a dessicator until experiments were started.

2. Preparation of Corrosive Solution

The minimum solution volume-to-specimen area ratio was 20 ml/cm2 of specimen surface, as recommended in ASTM A G31-72. The total solution volume was therefore fixed at 150 ml. The selected solvents were CH3OH and PE.

a) Hg Solution : Since the solubility of elemental Hg in cyclohexane is about 3 ppm., 5 mg of Hg was thus weighed and dissolved in 1 liter of cyclohexane.

b) HgCl2 Solution : The selected solvents were used to dissolve HgCl2. The HgCl2 concentration used was 150 ppm.

c) DMM Solution : The same selected solvents were used to dissolve DMM at concentrations of 0, 50, 100, 150, 200 and 250 ppm.

d). DMM + HCl Solution : DMM concentrations, as set in the method 2 c), were carried out, and 150 ppm of conc. HCl was added into each flask containing either methanol or PE. The concentrated HCl was a 37% by wt. solution, so introduction of HCl could thus not be separated from that of water. The volumes of conc. HCl and DMM required to give 150 ppm concentration in 150 ml is on the order of 0.05 ml, therefore, the total solution volume did not change significantly.

e) DMM + H2S Solution : Saturated H2S solution was prepared by passing H2S gas, which was synthesized from the reaction of conc. H2SO4 and Na2S powder, through the solvent for 1 h. at ambient temperature. In case of CH3OH, 1000 mg of H2S is soluble in 943 ml of CH3OH15. The solubility of H2S in diethyl ether is 6.86 x 10-4 mol/g while that of H2S in PE is 3.70 x 10-4 mol/g 16. Thus, in the case of PE, H2S gas was first dissolved into 10 ml of diethyl ether to saturation, then 140 ml of PE was added. The following corrosive solutions were prepared and studied using the method of  ref. 16:

- CH3OH or PE + saturated H2S

- CH3OH or PE + saturated H2S + 200 ppm DMM

- CH3OH + 300 ppm H2S solution + 200 ppm DMM

- CH3OH or PE + 150 ppm HCl

- CH3OH or PE + 150 ppm HCl + 200 ppm DMM

3. Procedure

One pair of specimens was immersed in an Erlenmyer flask containing corrosive solution. The flask was closed loosely by a stopper to release gas produced during the corrosion process and prevent the suppression of the reaction. Temperatures studied were varied from -10 to 70 °C with an increment of 20 ° C. At low temperatures, at or below 0 °C, the flask was stored in a refrigerator, whereas at higher temperatures a temperature controlled bath was used. Duration of exposure depended on temperature. At low temperatures, the corrosion process took a long time, 700-1,000 h. while at ambient and higher temperatures, the reaction took place very rapidly, 74.6 mdd, and needed to be observed at all time.

4. Specimen Cleaning after the Exposure

The cleaning process was done by following the methods of ASTM G1-72 (Reapproved 1979). Procedures were varied depending on type of metal being cleaned, as follows;

a) Aluminium Specimen Cleaning : Coupons were cleaned as well as possible with a plastic knife. Oily or greasy deposits were removed by soaking in trichloroethylene followed by the cleaning solution containing chromic acid, phosphoric acid, and water at 80 ° C for 25 min. The coupons were then rinsed with distilled water, isopropanol, and benzene. Finally, they were dried between paper towels and placed in a dessicator for 1 h. before weighing.

b) Carbon Steel Specimen Cleaning : After cleaning the coupons with plastic knife and soaking in trichloroethylene, remaining corrosion products needed to be removed by a bristle brush. They were then immersed in Clarke’s solution at room temperature for 25 min. The specimens were rinsed with water followed by isopropanol and dried between paper towels followed by warm air drying.

Results and Discussion

The study concentrated on the surface appearances of the corroded aluminium and carbon steel specimens and the effect of various factors on corrosion. The surface appearance was identified using scanning electron microscope.

a) Hg Solution : Figures 1 and 2 corresponding to the effect of Hg solution on the surface of carbon steel and aluminium specimens, respectively, show how Hg could corrode both samples and how important DMM was to the system containing DMM.

Figure 1 SEM photograph showing uniform corrosion of carbon steel in cyclohexane containing Hg (click on the image to enlarge it, press Back to return to the paper)

Figure 2 SEM photograph showing pitting corrosion of aluminium in cyclohexane containing Hg 

The corrosion occurred via the amalgamation of aluminium (eq.1)7. The amalgam is generally weaker than aluminium itself, thus easier to be attacked by water or CH3OH, see eq.2, and Hg generated would circulate to attack the aluminium again.

Hg + Al AlHg (1)

2AlHg + 6H2O 2Al(OH)3 + 2Hg + 3H2 (2)

The results appeared that at ambient temperature, the corrosion rates were 236.5 and 0.5 mdd (milligram/sq.decimeter/day) for carbon steel and aluminium, respectively. There were a lot of small pores distributed uniformly over the whole surface area of carbon steel specimen while a big black hole which seemed to be in the form of pitting appeared on the edge of aluminium specimen.

b) HgCl2 Solution : This is another severely corrosive agent15. Equation 3 shows the decomposition of HgCl2 to elemental Hg when reacting with aluminium.

2Al + 3HgCl2 2AlCl3 + 3Hg (3)

It was found that for the reaction in CH3OH system aluminium corroded very severely whereas in the PE solution, no reaction took place. The corrosion rates conducted in the CH3OH and PE solution were 1,179.5 and 0.015 mdd, respectively. This is due to the fact that HgCl2 is very soluble in either water or alcohol as methanol or ethanol, but not in non-polar solvents. PE is thus not a good solvent to dissolve HgCl2. However, the solvent PE used in the experiment contained 0.02% of water, as a result, very little reaction between aluminium and HgCl2 occurred in the PE system.

c) DMM Solution : In the flasks containing blank PE and MeOH solutions, after being immersed for 960 h. at ambient or 70 °C, specimens showed little corrosion, about 2 mg only, which is most likely due to scratch caused during the cleaning process, since the corrosion measurement after conducting the blank cleaning also resulted in almost the same weight loss, 2 mg. However, those immersed in the solutions containing DMM (see Figures 3 and 4 ) showed similar corrosion patterns to those immersed in the solution of Hg in cyclohexane, in that carbon steel and aluminium gave uniform and pitting corrosion, respectively. The higher the DMM concentration, the more pores or pits were observed on the specimens.

Figure 3 SEM photograph showing uniform corrosion of carbon steel in CH3OH + DMM

Figure 4 SEM photograph showing pitting corrosion of Aluminium in CH3OH + DMM

Interestingly, DMM in MeOH seemed to be more corrosive than DMM in PE. The discoveries of Skinner12 and Bass13 could explain this phenomenon. The C-Hg bond of DMM is weak and can be easily broken by either photolysis or thermal dissociation in the presence of a hydrogen donor, such as alcohol, as can be seen from eq. 4. The product is Hg metal which is a very corrosive species towards metals. PE is a non-polar solvent, and obviously not a hydrogen donor. Hg metal could thus not be formed. Corrosion on the specimen surface, however, could come from the presence of trace water in PE. 

CH3-Hg-CH3 + MeOH 2 CH4 + Hg (4)

The relationships of the corrosion rates versus temperature are illustrated in Figures 5 (for carbon steel) and 6 (for aluminium) for both MeOH and PE systems with and without DMM. The overall corrosion rates of carbon steel specimens were higher than those of aluminium specimens. Corrosion rates in solutions containing DMM increased significantly at temperatures above 30° and 50 °C for carbon steel and aluminium specimens, respectively, meaning that corrosion was a thermally activated process, as discovered by Skinner and Bass. 

It can also be seen in the case of the PE system, which contains trace amounts of water. Water could then serve as a proton donor to DMM, as shown in equation 5.

    (5)

Especially, at higher temperature, water molecules generate proton easier and donate proton to DMM faster. As a result, the decomposition of DMM to mercury metal, which is the cause of specimen corrosion, became easier.

Figure 5 Effect of temperature on the corrosion rate of carbon steel 

Figure 6 Effect of temperature on the corrosion rate of aluminium 

As described previously, organomercury could react with mineral acids to form mercury salts which are also corrosive compounds14, for example, forming mercuric chloride from the reaction of organomercury with hydrochloric acid.

Since natural gas also contains CO2, which could convert to carbonic acid in the presence of water, and H2S, the following studies were therefore undertaken to determine the effect of acid mixed in DMM solution on the corrosion rate.

As for the time dependence of the corrosion process, the experiment was carried out for aluminium in the CH3OH + 150 ppm HCl system, with and without DMM. The results are shown in Figure 7. It is obvious that DMM did not play the most important role as the exposure time increased.

Figure 7 Weight loss of aluminium specimen in HCl solution with and without DMM 

d) DMM + HCl Solution : After adding 150 ppm HCl into the flasks containing various concentrations of DMM (0 to 250 ppm), the corrosion results were rather frightening, especially with the systems having CH3OH as solvent, which produced a much more corrosive reaction than those with PE as the solvent. The reason is that HCl can not be easily dissolved in PE as compared to CH3OH. However, for both carbon steel (Figure 8) and aluminium (Figure 9), the corrosion rate increased with increasing DMM concentration. Moreover, the corrosion appearance of aluminium specimens seemed to be more uniform than pitting since it was now distributed over the whole surface area whereas carbon steel still showed uniform appearance.

The temperature dependence of corrosion rates of solutions containing only HCl and HCl with DMM were different, as seen in Figures 10 and 11 for carbon steel and aluminium, respectively. In the case of corrosion by the solutions without DMM and having CH3OH as solvent, the rates were obviously much lower than the ones with DMM after the temperatures went up beyond 30 (for carbon steel) and 50°C (for aluminium). This is simply because the DMM + HCl solutions contained not only corrosive HCl, mercury metal, but also mercuric chloride produced by the reaction of DMM/CH3OH and DMM/HCl as discovered by Skinner/Bass12-13 and Dessy14, respectively. For the case of PE as the solvent, the rates were much lower, but still higher than the ones without HCl. This phenomenon could be explained by the poor solubility of HCl in PE in comparison to CH3OH. However, when temperature was elevated, the solubility was increased from 1.53 x 10-3 to 9.83 x 10-3 mole/l.

Figure 8 SEM photograph showing uniform corrosion of carbon steel in CH3OH + DMM + HCl

Figure 9 SEM photograph showing uniform corrosion of aluminium in CH3OH + DMM + HCl

Figure 10 Comparison of corrosion rate of carbon steel in HCl solutions with and without DMM

Figure 11 Comparison of corrosion rate of aluminium in HCl solutions with and without DMM

e). DMM+ H2S Solution : Since H2S is a component mixed together with elemental Hg and Hg derivatives in natural gas, it was substituted in place of HCl in this study. Only aluminium specimens were studied and the results are listed in Table 2.

Table 2 : Comparison of corrosion rates of aluminium in CH3OH/PE + H2S/HCl + DMM solutions at ambient temperature for 130 h.

Corrosive Solutions

Corrosion Rate

mdd

mpy

CH3OH + saturated H2S

3.7

1.8

CH3OH + 150 ppm HCl

70.8

34.5

CH3OH + 150 ppm HCl + 200 ppm DMM

2,675.9

1,304.0

CH3OH + saturated H2S + 200 ppm DMM

3,516.6

1,713.6

CH3OH + 300 ppm H2S + 200 ppm DMM

2,603.7

1,283.3

PE + saturated H2S

0.05

0.02

PE + 150 ppm HCl

3.83

1.87

PE + 150 ppm HCl + 200 ppm DMM

5.40

2.63

PE + saturated ppm H2S + 200 ppm DMM

0.82

0.40

Note : mpy is mils (0.001 inch) per year

As can be seen, the corrosion in the solutions containing both acid and DMM increased drastically (Figure 12). Both acids have the same potential to cause corrosion. The rates were extremely high when DMM was present together with the acids. For example, CH3OH + 150 ppm HCl + 200 ppm DMM corroded the specimens approximately 2675.9 mdd, which was 40 times more than that corroded by the solution without DMM. The CH3OH + saturated H2S + 200 ppm DMM solution was severe as well. The rate was almost 100 times faster than the one without DMM.

Figure 12 : SEM photograph showing aluminium corroded by the CH3OH + saturated H2S + 200 ppm DMM solution at ambient temperature for 130 h.

As expected, when using PE instead of CH3OH, the corrosion rates were much lower. The system of PE + 300 ppm H2S + 200 ppm DMM also gave a lower rate than that of PE + 150 ppm HCl + 200 ppm DMM. This is due to the content of water in HCl since HCl used in the system was 37% of HCl in water while H2S used was in the form of gas purged into the PE solvent directly. The reactivity between H2S or HCl and DMM was then low. However, in gas phase as in natural gas, the gas molecules move around easier and faster, resulting in higher chance for those molecules to collide with each other. The reactivity between H2S and DMM should thus be better and the corrosion rate would be higher.

Conclusions

Based on the results of this study we find that organomercury compounds found in natural gas catalyze the corrosion of aluminium and steel components of gas pipelines and equipment in the presence of trace amounts of water. Decomposition of DMM leads to the formation of elemental mercury, which catalyzes corrosion by water. Furthermore, in the presence of small amounts of HCl or H2S the corrosive action of DMM is increased by several orders of magnitude and can lead to catastrophic failure in a relatively short time. These observations suggest that for natural gasses that contain mercury, steps should be taken to determine the presence or absence of organomercury compounds and eliminate them if possible.

Acknowledgements

This study was supported by the Royal Thai Government Research Fund. Deep appreciation goes to Professor Erdogan Gulari, Faculty of Engineering, University of Michigan, Ann Arbor, for his helpful suggestions, discussions and manuscript proof-reading.

References

1. Franken and Hubber, AIChE Spring National Meeting, Houston, April 1991, paper 27a

2. English, J.J., ibid, Houston, April 1989, paper 74a

3. Bingham, M.D., "Field Detection and Implications of Mercury in Natural Gas", SPE Production Engineering, May 1990

4. Haselden, G.G., "The Challenge of LNG in the 1980’s", Mech.Eng. 103, 46-53 (1981)

5. Kinney, G.T., "Skikda LNG’s Problem", Oil and Gas J. September 15, 1975

6. Leeper, J.E., "Mercury LNG’s Problem", Hydrocarbon Processing, 237- 40 (1960)

7. Phannenstiel, L.L., Mekialey, C., and Sorensen, J.S., "Mercury in Ntural Gas", paper PAP76-T-12 presented at the 1976 American Gas Assn. Operation Section Transmission Conference, Las Vegus, May 3-5, 1976

8. Osborne, S.P., Applied Spectroscopy 4, 1044 (1990)

9. Cappon, C.J. and Smith, J.C., Analytical Chemistry 49, 365 (1977)

10. Yamada, M. and Sato, K., "Mercury Removal from Natural Gas Condensate", Akio Furuta Kin-ura Research Center, JGC CORPORATION, 1992

11. McAuliffe, C.A., The Chemistry of Mercury, The Macmillan Press Ltd., London, 191-223, 1977

12. Skinner, H.A., Adv. Organometallic Chem. 2, 46 (1964)

13. Bass, K.C., Organometal Chem.Rev. 1, 391 (1966)

14. Dessy, R.E., Reynolds, G.F., and Kim, J.Y., J.Am.Chem.Soc. 81, 2683- 2688 (1959)

15. Budavari, S., Merck Index, 11th edition, Merck & Co., New Jersey, 512, 761, 1989

16. Kirk-Othmer, Encyclopedia of Chemical Technology Vol. 23, pg. 276, 4th Edition, John Wiley&Sons, 1997.