Effect of on Engineering Materials Used in Ammonia Plants

Inappropriate combinations of moisture, alloy and mercury can accelerate corrosion and form explosive compounds. Are there ways of preventing these potential hazards?

S.Mark Wilhelm Cortest Laboratories, Inc., Cypress, TX 77429

INTRODUCTION Secondly, concentration of mercury in LNG and ethylene plants has Elemental mercury (Hg) is becoming presented a range of materials an increasingly more prevalent related problems having to do with contaminant to natural gas feed aluminum. Lastly, reaction of stocks due to its production from mercury with ammonia (or NH3 various reservoirs around the world. precursors) can deposit potentially Table 1 shows the origin of gas explosive compounds in NH3 process having mercury contamination from a equipment. variety of locations(1>. This paper discusses the following In general, mercury contamination of points relating to mercury and its gas streams does not cause major influence on gas processing and use: operational problems for gas production, transportation or 1. Origin and composition of processing, at least from the mercury and mercury compounds standpoint of engineering. There in natural gas. are, however, environmental concerns from end use"'. Environmental issues Materials degradation will not be addressed in the present mechanisms. discussion. 3. Overview of mercury related Several situations have come to equipment failures. notice since 1975 that cause concern for the gas industry as they relate 4. Mercury - compounds. to mercury contaminated gas. One such problem is that minuscule 5. Recommendations for ammonia amounts of mercury in gas can induce plant operations. liquid metal cracking of a small group of susceptible alloys. The Mercury in Natural Gas problem has been manifested in the brittle failure of values and Table 1, as mentioned, provides the specialty components in production concentrations of mercury found in equipment. natural gas produced from selected

20 locations'1'. Mercury is scarce in 2. corrosion domestic (USA) supplies but has been detected in sizable proportions in 3. liquid metal embrittlement gas from North Germany, Algeria and Indonesia. The predominant form of 4. galvanic corrosion mercury is elemental; organic complexes account for less than 2 Amalgamation percent of produced mercury. Amalgamation is the process by which Sampling and detection of mercury is mercury forms liquid solutions with an inexact science and is various metals, primarily aluminum, considerably difficult with tin, gold, silver and zinc. Of inexperienced personnel. There can these, only aluminum has mechanical be, therefore, large uncertainties significance. The affinity of in measured concentrations aluminum for mercury is mediated by especially in liquid samples. The the oxide (A12O3) that protects the theoretical maximum amount of aluminum surface. Figure 2 mercury in gas is estimated by its illustrates the typical situation in vapor pressure at the temperature of which mercury contacts the aluminum the reservoir (see Figure 1). In all oxide surface layer protecting the known situations, the actual underlying metal. The oxide on measured concentration is less (by aluminum is not homogenous and at least an order of magnitude) than contains numerous defects. In the theoretical maximum deprived general, mercury lacks the ability; from the vapor pressure. This fact because of surface tension, to suggests that liquid mercury is not diffuse through cracks or defects a coexisting phase in gas reservoirs and hence cannot reach the and offers some reassurance that underlying metal. This lack of elemental mercury particulates will wettability, however, can be not be present in produced gas. mitigated by thermal or mechanical stresses, by chemical environments Mercury does not form stable and by temperature. complexes with most hydrocarbon derivatives and is not expected, nor In situations in which Hg can breach found, to exist as an organometallic the aluminum oxide and wet the species in any significant underlying surface, the rate of quantities. amalgamation depends in a major way on metallurgical (microstructural) Mercury adsorbs readily on condition. It is observed, for production and transmission instance, that mercury amalgamates equipment surfaces and, therefore, selectively with weldments and more its gas phase concentration is rapidly with particular alloys. location dependent, dropping off Table 2<1) provides a list of with distance from the wellhead. aluminum alloys and the kinetic Similarly, mercury has a significant degree to which they amalgamate with affinity for liquid hydrocarbon and Hg. partitions to the liquid phase in condensate separators, sweetening The primary manifestation of equipment, etc. The concentration of amalgamation is loss of mechanical mercury in coexisting hydrocarbon strength in weldments. Amalgamation liquid/methane can be estimated as does not require stress nor does it shown in Figure 1. require mediation by a conducting electrolyte to occur. MATERIALS DEGRADATION MECHANISMS Amalgam Corrosion Mercury can degrade materials by four basic mechanisms: Amalgam corrosion is the combined action of Hg and moisture on 1. amalgamation susceptible materials, primarily aluminum and tin. The difference

21 between this mode of attack and surface contamination. The limiting simple amalgamation is that the crack velocity is approximately 100 corrosion process propagates with cm/s which is reached due to the minuscule amounts of mercury. The limit of liquid diffusion of mercury reaction scheme is: (not surface diffusion or vapor diffusion). In Figure 4, the Hg + Al Hg(Al) amalgam limiting stress intensity factor for [1] mercury LME of aluminum is shown. For weldments it can approach 5 ksi Hg(Al) + 6 H2O 3H2 in1/z which translates into a Hg [2] critical defect size of 10~* mm. LME is not limited to aluminum but Hg + Al -» Hg(Al) [1] can affect high strength steels as well. Susceptible materials include etc. 4140 RC 35-40 at elevated temperature and precipitation Amalgam corrosion regenerates the hardened stainless steels also reactant and hence is self- having high strength and high propagating. If sufficient moisture hardness. and mercury are present, aluminum structural components can be Galvanic Corrosion penetrated fairly rapidly. The rate of attack is mass transfer limited Galvanic corrosion caused by mercury but does not proceed as rapidly as deposits in normally corrosive liquid metal embrittlement (LME) environments has been observed for discussed below. Amalgam corrosion steels. Mercury serves to accelerate is selective to the same alloys as acid dissolution and to aggravate simple amalgamation, however, it can localized corrosion such as pitting. affect all aluminum alloys to some The situations in which mercury degree. deposits contact steel are found in gas/water separators in gas Liquid Metal Embrittlement (LME) production and when mercury from gauges contaminates process piping Liquid metal embrittlement by in chemical plants. mercury is distinct from amalgamation in that it produces MERCURY RELATED FAILURES IN GAS rapid brittle fracture and affects a PROCESSING EQUIPMENT much broader range of materials (aluminum, nickel-copper alloys, LME of Aluminum Cold Box Pipina(5) brasses, copper alloys, tin alloys, some stainless steels). The Ethylene plant cold box piping (see mechanism of LME involves liquid Figure 5) has experienced LME due to diffusion of mercury atoms in grain contact of elemental mercury with boundaries. Cracks usually initiate aluminum alloy 5083 weldments. The and propagate via the grain mercury originated in Algerian boundaries. LME is distinct from feedstocks (LNG) that had been stress corrosion cracking in that no processed through the cold box for a purely electrochemical processes are considerable period of time. The involved. mercury contamination was detected ultrasonically and radiographically. The unusual aspects of LME, as Failure occurred upon start-up after opposed to other fracture processes, a plant shutdown at weldments in are that the crack propagation rate cold box piping. can be exceedingly fast and the stress intensity required for crack Many ethylene plants have now propagation can be very low. Figure installed sulfur-imprégnated carbon 3 illustrates the rapidity with filters to adsorb mercury in inlet which aluminum cracks propagate in gas. Methods to clean existing smooth tensile specimens having equipment that may be contaminated

22 (1) have not been developed as yet. Corrosion of Stainless Steel Detection and location of cracks is very difficult. Acoustic emission 316 L stainless was severely shows some promise, but accurate corroded when subjected to a mixture procedures to measure existing of CH4, CO2, H2S water and mercury. damage are still in the Both water and mercury were developmental stage. required. Corrosion took the form of pitting and crevice attack. The LME of Monel 400 Valve Stems and affected equipment was a laboratory Springs research pressure vessel. Valve springs and valve stems of CHEMICAL REACTIONS OF MERCURY IN Monel 400 were found to experience AMMONIA SYNTHESIS LME upon exposure to raw natural gas containing approximately 100 ug/m3 Mercury - nitrogen compounds are elemental mercury. No condensing notoriously explosive and include: mechanisms (i.e. cold temperatures) were present. Mercury was delivered mercury to the susceptible pieces by Hg2N3 adsorption from the gas and/or condensate. mercury (halide) nitride Hg2 (X) N Monel 400 has a composition of where X = Cl", Br", T approximately 66 percent Ni, 32 percent copper with minor amounts of mercury hydroxynitride iron, manganese and silicon. Copper- Hg2NOH nickel alloys have well known susceptibility to LME by several mercury oxynitride liquid metals. The degree and rate (Hg2N)20 of attack increase with temperature. The unusual aspect of the Monel Mercury nitride (Hg2N3) is the most valve part failures is that common form of mercury-nitrogen extremely low mercury levels were compound. It forms routinely in sufficient to produce a cracking mercury manometers on vacuum systems failure. Failure occurred at that handle NH3 gas and its approximately 1000 C. Evidently detonation is the source of several mercury adsorbed onto the metal laboratory accidents. The mechanism surface and this amount was by which Hg2N3 forms is complex, but sufficient to produce LME. is known to require oxidation of mercury. In vacuum systems, moisture Amalgam Corrosion of Cryogenic Heat is involved Exchangers in Skikda (Algeria) LNG Plant(6) Hg H2O HgO + H2 [3]

Aluminum heat exchangers (A65 French 4 HgO + 6 NH, 2Hg2N3 + 4 H2O -H 5 Standard, 6061 American Standard) H, [4] experienced amalgam corrosion upon exposure to gas containing 0.001 to The gas phase reaction of mercury or 0.65 ug/Nmr mercury. Corrosion mercuric oxide with ammonia has not occurred at a temperature of O C or been studied. Presumably mercuric slightly higher. The problem at oxide would be formed during syngas Skikda was due to circumstances in production and carried through to which the cycle gas stream was the catalyst. Contact with nitrogen between O C and the dew point of compounds at elevated temperature water. This occurred only during could then generate the nitride. deriming or plant shut down. The deposition of liquid water along RECOMMENDATIONS FOR AMMONIA PLANT with mercury on heat exchanger OPERATORS surfaces was required for the corrosion to take place. Mercury exists and will continue to

23 exist in natural gas feedstocks. compounds are explosive. The Because of the difficulties in kinetics of formation, the diversity sampling and analysis, the exact of species (halide, oxy, hydroxy, concentration of mercury in gas etc.), the chemical mechanism by feedstocks can be uncertain. As new which they form and the ultimate and deeper fields are brought on disposition of minuscule amounts of stream, the chances for mercury, at Hg2N3 in ammonia plant should be higher concentrations than investigated. There do not exist, at previously encountered, to be present, sufficient technical data processed into syngas and to assess the possibility of hazards subsequently used in ammonia associated with mercury-nitrogen production are finite. Given the compounds. background of the present discussion, several recommendations ACKNOWLEDGEMENTS are to be noted: Helpful discussions with Mr. J. J. 1. Ammonia plant operators should English of Oxychem are acknowledged be aware of the mercury content with appreciation. Some of the of gas feedstocks. This may technical data provided in this require special procedures to paper were developed in conjunction analyze gas streams for with projects conducted by Mr. Dale mercury. Bingham(7) has recently Mclntyre (now with Aramco, formerly reviewed mercury detection and with Cortest) for Cain Chemical (now measurement techniques in Oxychem) and E.I. DuPont. natural gas production and process streams. Furthermore, REFERENCES operators should be aware of equipment that can remove 1. W.W. Bodle and R. Serauskas; mercury if required. In "Considerations for Mercury in LNG ethylene plants, mercury is Operations"; IGT, 6th International removed by sulfur impregnated Conference on Liquified Natural Gas; carbon which can be used at Kyoto, Japan; April 1980. pressures up to 1500 psi. Removal efficiencies are such 2. R. Härtung and B.D. Dinman, that effluent concentrations of "Environmental Mercury less than 0. l /ig/™3 can be Contamination"; Proceedings of the achieved. Calgon Corporation International Conference on Mercury, can be contacted for details. Ann Arbor Science Publishers, Inc., 1970. 2. Materials incompatibilities should be avoided. Primarily 3. P. Gordon; "Metal Induced this means a close examination Embrittlement of Metals, An of the use of aluminum and Evaluation of Embritter Transport nickel-copper alloys. Generally Mechanism"; Metallurgical speaking, aluminum is not found Transactions A, 9A, 267, 1978. as a pressure bearing component in any part of conventional 4. D.R. Mclntyre, J.J. English, G. ammonia plants. Copper-nickel Kobrin; "Mercury Attack of Ethylene alloys are not common either, Plant Alloys"; NACE Corrosion/89, however, valve stems, seats and Paper No. 106; April, 1989. springs occasionally find their way into process equipment; 5. E.P. Dahlberg; "Cracking of these should be avoided. Aluminum Heat Exchange Piping"; Mercury will not corrode or report to J.J. English, Cain crack common steels which is Chemical Company, June 11, 1987. encouraging for the vast majority of equipment. 6. J.E. Leeper; "Mercury Corrosion in LNG Plants"; Canadian Gas 3. Ammonia plant operators should Processors Association, Quarterly be aware that mercury-nitrogen Meeting, Calgary; September, 1980.

24 7. M.D. Bingham; "Field Detection and Implications of Mercury in Natural Gas»; SPE Petroleum Engineering; pp. 120; May, 1990.

S. Mark Wilhelm

SMOOTH TENSILE SPECIMEN, Hg CONTAMINATED. CRACK LENGTH 10mm

i 40 60 80 100 120 140 APPLIED STRESS - % OF .2% OFFSET YIELD STRESS Figures. Delayed failure time vs. applied stress for 2024 TO TEMPERATURE 'C aluminum embrlttiement by Hg at room Figure 1. Saturation concentrations of mercury In gases and temperature. In hydrocarbon liquids.

\ r WELD

10-

BASE OlAL

345678 91011121314151617 Al203 Ksi Jin"

Figure 4. Crack growth rates' at room temperature, Alloy Figure 2. Amalgamation. 5083.

25 Table 1. Elemental mercury concentrations In natural gas.

Location Algeria (Wellhead) 50

Algeria (pipeline entrance} 0.1

Algeria (Skikda plant inlet) 0.001

Groningen (wellhead} 180

Groningen (to pipeline) 12

North Germany (wellhead) 15 450

South Germany (wellhead) <0.1 0.3

South America 69 119

Far East (Pakistan) 3 20

DOHNSTREMI 5EPWW50N5 Far East 58 193 Figure 5. Ethylene plant cold box location of LME piping Far East 0.02 0.16 failure. Africa .(Angola) 0.3 130 Middle East (Iran) 1 9 JJ12, Eastern U.S. Pipeline 0.019 0.44 2 BARSl Midwestern U.S. Pipeline 0.001 -163-C *-D 0.10 North America 0.005 0.040

Sumatra, Indonesia 200 300

Table 2. Compositional Analysis and corrosion test results for aluminum alloy samples.

Kg Kn Dl Tf Si Cr Alloy (vt- (Wt- „«- (wt- EKt*nt of Designation *) *> "t *) "ir "i; *) corawlon* Y*9 Catastrophic,

5083 4.4 O.SB 0.04 0.24 Ho

0.1- M04n«ta, 300-R1S Tc»e« 1-5 0.2 O.T o.s non-dawlng

Y«a-Kod«rBt« Ï.1- 0.15 potentially

2.1- 0.3 l!o e!i ... 0.7 ...... BO "turn" indicate» that cat«!trophic c -Ho- indicate» that It wao not.

LOW-PRESSURE HIGH-PRESSURE CASING CASING Figure 6. Mixed refrigerant LNG plant, location of amalgam corrosion.

Georg Grossmann, BASF Ludwigshafen: What happens Grossmann: Thank you. You mentioned sulfur- to mercury in liquified natural gas after revaporization? You impregnated carbon as a means of removing mercury. Do told us that it normally is contained in the liquid product. you think plain activated carbon will do also? Wilhelm: It vaporizes. Wilhelm: My understanding is that it does, but it is not Grossmann: Will it evaporate together with methane and highly efficient It needs to be regenerated or replaced more will be in the methane output stream? frequently than the sulfur-impregnated carbon. The Wilhelm: Yes, sir. It does not remain with the liquid economics of regeneration or replacement dictates the use of phase. It vaporizes. the sulfur impregnated carbon. Grossmann: Do you know whether in the revaporization Gary Smith, Fire Chief in Watsonville: The fact that plant any measures are taken to remove mercury again? ammonia explodes in the refrigeration plant has confused Wilhelm : I do not know of any. everybody in figuring out why. Do you think that in a free

26 release, if a cloud filled up, the mercury could be a potential which we can make an educated guess as to what effects factor in adding to the explosiveness of an ammonia cloud? mercury might have. Wilhelm: I seriously doubt that the mercury in its Lastly, given a scenario where mercury forms explosive elemental form would affect the explosion of an ammonia compounds with ammonia in the synthesis loop, the refrigeration plant. It's the nitride that has the energy. questions then are: How does the mercury get through the Oxidation of mercury is not a highly exothermic reaction, so process to the synthesis loop, will it be concentrated in the it should not participate significantly in an explosion. ammonia separator, and will sufficient concentrations K. R. Krishnaswami, Madras Fertilizers, Madras, accumulate and then will it detonate? Here again, we don't India: The last three papers have referred to the mercury have sufficient information to really answer all of those effect in natural gas and consequent catastrophic failures. We questions. The initial reaction in looking at the process is have been in the ammonia plant business operating with that mercury can get to the synthesis loop and it can react natural gas feedstock for several decades. Now the mercury with the mercury. Whether or not it is going to be effect is being highlighted. Do you think there is any concentrated in the separator, I don't know. The quantities process failure through which mercury has leaked to the that are required are going to be highly selective depending synthesis loop, and in your case when the mercury formed on plants and feedstocks. complex with the ammonia? Were there any recent process A. Nielsen, Haldor Topsoe: Thank you for your very failures that have not been reported? Or is this a new important paper. I believe the first discussion of explosive phenomenon; we have to find out, since the mercury failures compounds formed from ammonia, mercury and traces of have not been highlighted very much and not very much water is by Michels, Dumoulin and Gerver of the Van der heard of before. Waals Laboratory in Amsterdam ( Ree Tra Chim Pays-Bas, The other question is about quantifying the presence of 76, pp. 5-12,1957). mercury. Limiting or maximum quantity of mercury that can According to these authors, the compound with the cause damage is also not clearly brought out. Is there not strongest explosive power is the lowest of the four exact scientific analysis to find out the mercury level in compounds in your list, influent natural gas? And are there any precautions that we containing a certain number of ammonia molecules in the can take to avoid such catastrophes? structure, which apparently is critical for obtaining the Wilhelm: My personal view is that mercury is a recent highest degree of explosive power. Also, the next lowest phenomenon in domestic plants because of the current gas compound from your list, formed from the lowest one by market. Gas that originates from remote locations is now reaction with water, is explosive, but to a lower degree. finding its way up to our shores. That is a recent A number of attacks by mercury on engineering materials phenomenon. Secondly, we know that mercury is more used in ammonia plants were reported hi the early volumes prevalent in deeper reservoirs, meaning that the likelihood of of Ammonia Plant Safety Technical Manual. All of these finding mercury has probably increased recently as we go to referred to the use of some hydrogen from mercury cells. In deeper depths. That's a long-term trend. The combination of Volule 1 of the Manual, R. F. Bollen of Dow Canada feedstock mobility and prevalence means that the presence of reported an attack on Admiralty brass tubes in a first-stage mercury in plants is likely to be a recent phenomenon. intercooler. In Volume 2, J. Clapperton of Columbia Given that fact, how likely is it that mercury actually Southern Chemical Corporation reported a ring failure in the participates hi the failures that we have seen historically and sixth stage of their syngas compressor, and H. A. Sommers are seeing now? This is more difficult to answer. Very little of Pennsalt Chemicals reported that, while he was with Olin research has been conducted on the effect of mercury on Mathieson at Niagara Falls, they at one time installed three normal failure mechanisms, i.e., the ammonia stress new primary compressors, of which the second and third corrosion cracking mechanism or galvanic corrosion of stage after coolers had brass tubes that had been damaged by stainless steel. So there is just very little information on mercury very quickly.

27