INFLUENCE OF PETROLEUM COKE SULPHUR CONTENT ON THE SODIUM SENSITIVITY OF CARBON ANODES Sheralyn M. Hume, Werner K. Fischer, Raymnd C. Perruchud R&D Carbn Ltd., P.O. Bx 157, 3960 SIERRE / SWITZERLAND James B. Metsn, Department f Chemistry University f Auckland, Private Bag, Auckland, New Zealand R. Terry K. Baker, Materials Research Labratry The Pennsylvania State University, University Park, PA 16802, USA ABSTBACI Carbn andes can cntain significant quantities f sdium depending n the raw materials and recycled butts cleanliness. The extent f the catalytic effect f the sdium n the ande reactivity, i.e. the "sdium sensitivity", is dependant n the cke nature. An investigatin f a range f petrleum ckes was undertaken t study the influence f the sulphur n the sdium sensitivity. Besides the standard characterisatin tests, a range f analytical methds was emplyed including Cntrlled Atmsphere Electrn Micrscpy (CAEM), X-ray Phtelectrn Spectrscpy (XPS), Secndary In Mass Spectrmetry (SIMS) and ThermgraVimetric Analysis (TGA). The results f these investigatins have been used t elucidate a mechanism t explain the sdium sensitivity. A shrt vide shwing the cntrasting behaviur f tw ckes at the micrscpic level will be shwn. INTRODUCTION The raw materials used fr the prductin f the dry recipe in prebaked ande plants are petrleum cke, crushed butts, green scrap and baked scrap. A binder, typically cal tar pitch, is added t cat the filler aggregate particles and ensure the cmpacted ande blck maintains it's structural frm. prblem f crrsin in the fractinating equipment due t chlride attack. Unfrtunately the added sdium ends up in the pitch and subsequently the ande. Spent ande butts are remved frm the cell, washed and crushed befre mixing with the pitch and cke t frm the ande. A large level f the impurities present in the ande is cncentrated in the butts. Butts may als cntain impregnated crylite (Na 3 AIF 6 ) and bath material (a mixture f crylite and alumina A1 2 0 3 ) depending n the efficiency f the washing prcedure. Dirty butts may cntain as much as 12000 ppm sdium. The extent f the catalytic effect f the sdium n the ande reactivity, i.e. the "sdium sensitivity", is dependant n the cke nature. R&D Carbn Ltd develped a bench scale prcedure t quantify this which was presented in Light Metals 19912. Figure 1 shws the influence f increasing sdium (dirty butts) cntent n the carbxy reactivity residue f full size andes made ut f tw different petrleum ckes. C02 REACTIVITY RESIDUE (%) 1 00 ~------~----~------~----~ ANODE L Carbn andes can cntain significant quantities f sdium depending n the raw materials and recycled butts cleanliness. Sdium can be inherent in crude il, r acquired during the drilling prcedure. During il refining and cking, the majrity f the metals are depsited in the cke. Further sdium can be depsited n the cke while decking the drum and after calcining if the purity f the water used fr the cutting and cling peratins is nt cntrlled. Prir t tar distillatin fr pitch prductin the feedstck is dsed with cntrlled amunts f either a sdium carbnate r sdium hydrxide slutin t neutralise ammnium chlride disslved in the tar water'. This eliminates the Figure 1. 70 ;-------~----_+----~~~---; 60 ~-----+------~----~----~ 0.5 1.0 1.5 2.0 (%) ANODE H BUTTS SODIUM CONTENT Influence f the butts sdium cntent n the carbxy reactivity residue f andes.
There is a strng indicatin that the sensitivity f the ckes t dirty butts additin (i.e. their sdium sensitivity) is different. The phtgraph in figure 2 illustrates the large differences that can ccur in andes made with ckes having a similar granular carbxy reactivity. R&D ICOKE SODIUM SENSITIVITYI,.,,,. "" BENCH SCALE ELECTRODES, COKE : M, LOW SULPHUR PITCH : CHEMAPOL 8 70/92 RECIPE,::UCaONTENT. : STANDARD LINE _. _(ppm) 1000 _ ~- 500 - - -. - C::.6REACTIVITY RESIDUE.\%) RANGE OF COKES RECIPES :!:1 Jl (60 %) A 0,00 " N B 0,05 0. Ne C O.1 0 '-. Na 0,1[, -.. Na C02.~=-Y ~U~ \") 12 - - - - 8 -. 4 _. - 1' - I C:2g:E_AC: VITY LOSS(") A B C 500 100 1 00" llfi1n.. 80 - - - - 60 70-1 2-8 -. ~ 4 A 8 C 0 ODIU ITI IT Figure 3. Bench Scale results fr lw sulphur cke M. Figure 2. The carbxy reactivity residues f high and lw sdium sensitivity ckes. A cmparisn f the results f many such investigatins led us t believe that the petrleum cke sulphur cntent had an imprtant influence n the sdium sensitivity (ther imprtant parameters include the cke prsity, calcinatin degree and macrstructure). An investigatin f tw petrleum ckes with high r lw sulphur but therwise similar characteristics was undertaken t study the influence f the sulphur cntent. Besides the standard characterisatin tests, a range f analy1ical methds were emplyed. R&D ;,,, I "~III. J COKE SODIUM SENSITIVITY BENCH SCALE ELECTRODES COKE : S, HIGH SULPHUR RECIPES A : 0,00,., Na RANGE OF B : 0,05 J. Na PITCH : CHEMAPOL 8 70/92 :!::L-. (68 %.) c : 0,10 % Na RECIPE : STANDARD LINE :::i&i!f 0; 0,15 ". N n COKES 12 --l=- _ 8 'It- =t=:j-=! 600 - - ~ 4 -.- - BCD C02 REACTIVITY RESIDUE (%) C02 REACTIVITY LOSS (%) 90 r- 80 12 70 ~ 1--.~ 1- - - 60 4 A B C D I 16 'H---t---j---l I:):~:=r=:.tl BENCH SCALE PROCEDURE Bench scale electrdes cntaining increasing levels f sdium cntaminatin were prepared accrding t the prcedure f Fischer and Perruchud 2. The bench scale results fr typical lw (labelled M) and high (labelled S) sulphur ckes are shwn in figures 3 and 4 respectively. Figure 4. Bench scale results fr high sulphur cke S. It is evident that the sensitivity t sdium additins is much greater fr cke M, ie the cataly1ic activity f the sdium is much greater in the absence f sulphur.
ThermQravimetric Analysis QGN A Stantn Redcrft fully autmated thermbalance was used t measure the carbxy reactivity f the calcined petrleum ckes and crrespnding crylite cntaminated binder matrix electrdes. Binder matrix electrdes are prepared using nly the dust fractin in the dry aggregate 3. In the case f analyzing small cntaminated samples, using binder matrix electrdes enabled a mre hmgeneus cntaminatin and therefre mre reprducible results. A 100mg sample f crushed material (grain size 1-1.4mm) was heated at a rate f 1 C/min in an atmsphere f flwing (21 mumin) carbn dixide frm 550 C until a 90% weight lss was btained. 10.-,-~--_- --~--~~ Cke S Binder Matrix S Cke M ~ 1 Binder Matrix M E ~.. 1ii CI: 0.1 '. Figure 6. <~\ '..-""\ Seary Ins Pr1mary In'S\ Schematic diagram f Secndary In Mass Spectrmetry (SIMS). In electrdes made frm petrleum cke and cal tar pitch, element maps can be used t study the distributin f impurities within the electrde. As vanadium is present predminantly in petrleum cke and nt in pitch, and zinc is present predminantly in pitch but nt petrleum cke, a cmparisn f the sdium distributin with the distributins f these elements can prvide infrmatin n the mbility f sdium. Figure 5. 0.01 ---I, \- ' ---'-_, 1200 11 00 1000 900 Temperature 0 C Arrhenius plts f carbxy reactivity. The results are presented in the frm f Arrhenius plts in figure 5. In the case f the lw sulphur cke M, the granular cke material shws a lw reactivity and slw reactin rate. When electrdes are prepared frm this cke cntaminated with crylite, the reactin prfile changes significantly. The nset f reactin is apprximately 100 C lwer and therefre the rate f reactin is faster at any given temperature. This difference is attributed t catalysis f the carbxy attack by sdium. The situatin fr the high sulphur cke S hwever is quite different. The reactin prfile f this cke is relatively unchanged when electrdes cntaminated with crylite are prepared. The nset f gasificatin is at the same temperature (-BOO C) and the reactin rate at any particular temperature is similar. Here it appears that the catalytic activity f the sdium has been inhibited. Figure 7 shws element distributin maps fr baked bench scale electrdes prepared frm ckes lw and high in sulphur. Fr the lw sulphur material, the sdium distributin is thrughut the entire material, shwing n preference fr the cke r pitch phase and suggesting that the material is highly mbile. In the high sulphur material the sdium distributin is cincident with the petrleum cke phase. This is suggestive f a strng interactin between the sdium and a cke impurity (sulphur) which binds the sdium t the cke and thus prevents further migratin and catalysis. Secndary In Mass Spectrmetry (SIMS) SIMS 4 is an analytical technique that can be used t characterise the surface and near surface regin f slids. The technique uses a beam f energetic (0.5-20 kev) primary ins t sputter the sample surface, prducing inised secndary particles that are detected using a mass spectrmeter. In additin an image can be frmed using a fcused in beam (in a manner analgus t that f the electrn beam in SEM) and in this way element specific maps r images can be generated. A schematic verview f the technique is shwn in figure 6. Figure 7A. SIMS element distributin maps (diameter 150pm) fr electrde cntaining lw sulphur cke M.
Sectins f the electrdes were carefully grund dwn t sizes f the rder f 100f.1m using a dry grinding prcedure with a diamnd saw t avid cntaminatin f the specimen. These fragments were then ultrasnically dispersed in n butanl. A drp f this suspensin was applied t a transmissin specimen f single crystal graphite supprt munted n a stainless steel micrscpe grid. This cmbinatin was placed in the heating stage and inserted int the envirnment cell lcated in the specimen chamber f the micrscpe. Specimens were then reacted in the presence f 0.3 Trr f carbn dixide. Tw series f transmissin electrn micrgraphs taken during the curse f a reactin are shwn in figures 9 and 10 fr lw and high sulphur cke respectively. The behaviur f the tw materials is significantly different. 10 min 30 min Figure 7B. SIMS element distributin maps (diameter 150f.1m) fr electrde cntaining high sulphur cke S. Cntrlled Atmsphere Electrn Micl.QS~ (CAEM) Cntrlled atmsphere electrn micrscpy (CAEM)5 is an in situscanning transmissin electrn micrscpy technique. By a clever design adaptatin, it vercmes the principle limitatin f cnventinal transmissin electrn micrscpy, ie. a vacuum cnditin, and thus allws direct bservatin f the changes taking place in a transmissin specimen as it is heated in the presence f a gas. The advantage f such a technique is that the investigatr is nt required t remve the specimen frm the reactin envirnment ver perids f time t cllect cmparative micrgraphs, and s the assciated prblems f specimen cling, cntaminatin and subsequent relcatin f the regin f interest are vercme. A simple schematic diagram f the micrscpe is shwn in figure 8. 50 nm Transmissin electrn micruraplls f I w sulphur pelrnl 'Urn cke in CO. 10 min 30 min -~-~ electrn gun 70"", eperture \ -7 < 10 trr \ electrmagnetic lens system sample gas Inlet vecuum-, [~ \ ;.- gas l '.. sample \ \.. J _7-\'1~[---- in;:~~81~~ectln \ < 10 IO~ 70"", aperture Figure 8. Schematic f the Cntrlled Atmsphere Electrn Micrscpe. Figure 10. 50 nm Transmissin electrn micrgraphs fr high SUlphur cke in CO 2,
In the lw sulphur material, the catalytic attack f the sdium derived frm crylite is extremely high. Particle nucleatin and the assciated frmatin f tiny pits ccurs at an activatin temperature as lw as 200 C (depending n the heating rate). The nset f gasificatin ccurs fllwing nucleatin, ccurring by an edge recessin mde which is characteristic f a metal xide species, and cntinues until all the carbn has been cnsumed. INTENSITY (COUNTS) 3000 2000 LOW SULPHUR COKEM NalS In the presence f sulphur this catalytic activity is cmpletely inhibited. N catalytic activity is bserved and gasificatin des nt ccur befre a temperature f 860 C, which is cnsidered t be uncatalysed gasificatin. The absence f any visible catalyst signifies that the impurity is nt present as a metal xide but instead in a strngly bund spread state (which is invisible in TEM) and therefre unable t underg the catalytic xidatin/reductin cycle which is bserved with sdium species in the absence f sulphur. 1000 O +------r-----,------r---~ 1080 1075 1070 1065 ev INTENSITY (COUNTS) X-rav Phtelectrn Spectrscpy---LXESl. X-ray phtelectrn spectrscpy6 is a highly sensitive surface analysis technique used t determine the elemental cmpsitin f the utermst atmic layers f a slid and t btain detailed knwledge abut the chemical binding states, surface hmgeneity and the state f adsrbates. The basic prcess invlves the adsrptin f x-rays by an atm and the subsequent ejectin f cre r valence electrns. Infrmatin abut the elements present is extracted frm the binding energies f ejected electrns and the related shifts in these energies. HIGH SULPHUR COKE S 3000 Na1s 2000 1000 O +-----~----_r----_r----~ 1080 1075 1070 1065 ev Narrw windw scans f the Na1 sand S2s regins are shwn in figure 11. Figure 11A. XPS Na1 s narrw scans. A Na auger parameter can be calculated frm the binding energies frm the fllwing equatin: AP = 1254 - BE(NaKll) + BE(Na1 s) where AP = auger parameter BE = binding energy The chemical shifts and auger parameter are tabulated belw. Table 1 XPS data Cke S Cke M C1s 284.4 284.4 Na1s 1072.7 1072.2 NaKLL 263.9 263.7 AP 2062.8 2062.5 S2s 236.8 / 228.3 236.4 S2p 163.8 164.0 The value f the auger parameter and the shift in the S2p binding energy can be used t identify the chemical state. The Na peaks here are assigned as an xide r sulphate species, the sulphur peak at -236.6 shws the presence f sulphate. Of interest is the presence f the secnd sulphur peak in the high sulphur material which is assigned t elemental S r a carbn sulphide. This peak is nt present in the lw sulphur material, where all available sulphur appears t be in sulphate frm. It is pstulated that insufficient sulphur was present t bind all the sdium whereas in the high sulphur material, unbund sulphur is still available t neutralise the catalytic activity. It is believed that the neutralisatin takes place during the baking phase f the ande preparatin.
INTENSITY (COUNTS) 300 200 100 236.8 HIGH SULPHUR COKE S S2s O ~--~-----r----'-----.---~ 245 240 235 230 225 ev It can be pstulated that by frming a stable nn-mbile cmplex between sdium and sulphur, the sdium is prevented frm (a) diffusing Int the cke and (b) catalysing the reactin at active sites. In the case f sdium added (eg crylite), this wuld stp reactin almst entirely as the sdium species reacts preferentially with sulphur t frm a stable inactive species instead f xidatin f the sdium species t a cataly1ically active frm. Hwever if the sdium was already distributed thrughut the cke material in the xide frm (eg frm Na 2 C0 3 cntaminated cling water, then nly a reductin in activity wuld be expected (and is bserved) as the active catalyst is "pisned" by the sulphur during the redx cycle. INTENSITY (COUNTS) 300 200 100 236.4 LOW SULPHUR COKEM S2S 0 245 240 235 230 225 ev Figure 11 B. XPS S2s narrw scans. A MOPEL FOB TH E SOPIUM SENSITIVITY The general mechanism fr the cataly1ic gasificatin given belw was utlined by Mulijn and Kapteijn (1985) 7, taking int accunt all the pstulated mechanisms in the literature and published experimental kinetic and mlecular species data: where CO2 + MxOy P CO + MxOY+1 M x Oy 1 + C... MxOy + C(O) C(O)... CO M = a metal atm x,y,z = number f atms CONCLU SIONS The dependence f the sdium sensitivity n the cke nature has been cnsidered as a functin f the cke sulphur cntent. Frm a bench scale prcedure the different sensitivity t sdium attack was established. Thermgravimetric analysis shwed this t be due t a change in the reactin mechanism. SIMS element mapping shwed that during baking, the sdium was bund t the high sulphur cke grains and prevented frm migrating In the absence f sulphur a mre unifrm sdium distributin was fund, CAEM identified the cataly1ic species as a metal xide, prmting reactin thrugh nucleatin and edge recessin. This cataly1ic activity was arrested in the presence f sulphur. Finally XPS identified a sdium sulphate species, and when excess sulphur was present a secnd sulphide species. It is pstulated that the lwer sensitivity f high sulphur ckes is due t a binding by sulphur f the sdium species in an inactive frm preventing catalysis f the gasificatin reactin. It is believed that this neutralisatin takes place during the baking phase f electrde preparatin when the temperature is high enugh t allw sdium migratin. With the current trend in cke quality t higher sulphur levels, mre pressure is being put n smelters t reduce the quantity f sulphur in andes, primarily fr envirnmental reasns. Cnsequently cke prducers are increasingly being pressured t desulphurise their cker feedstck. This desulphurised cke is much mre sensitive t sdium additins and the excess carbn cnsumptin can increase dramatically if it is used in place f less sdium sensitive ckes where sdium cntaminatin is high. Hwever this cke can be used t make excellent quality andes if attentin is paid t the cleanliness f the recycled butts, An excellent review f butts cleaning systems was written by Rbert Tnti B
REFERENCES 1. D. Bradley (1986), ''Tar Distillatin Tday." In The Cke Oven Manager's Yearbk, p. 99. 2. R.C. Perruchud and W.K. Fischer (1991), "Determinatin f the Sdium Sensitivity f Petrleum Cke." Light Metals, AIME, New Orleans, p. 581. 3. M. Smith (1991), "An Evaluatin f the Binder Matrix in Prebaked Carbn Andes used fr Aluminium Prductin", PhD thesis, University f Auckland, New Zealand. 4. A. Benninghven, F.G. Rudenauer and H.W. Werner (1987), "Secndary In Mass Spectrmetry", Jhn Wiley & Sns. 5. R.T. K. Baker, "Factrs Cntrlling the Mde by which a Catalyst Operates in the Graphite-Oxygen Reactin", Carbn, Vl 24, N 6, pp 715-71 7, 1986. 6. D. Briggs and M. P. Seah (1983), "Practical Surface Analysis by Auger and X-ray Phtelectrn Spectrscpy", J. Wiley & Sns. 7. J.A. Mulijn and F. Kapteijn, "Catalytic Gasificatin", In "Carbn and Cal Gasificatin", Eds. J. Figuered and J.A. Mulijn, Prc. Nat Advanced Science Institute, Alvar Prtugal, pp 181-195, 1985. 8. R.T. Tnti (1989), "Imprving Ande Perfrrnance", Light Metal Age, December, p 12.