VOL. 5, NO. 6, JUNE 2010 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences ©2006-2010 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com THE NATURE AND EFFECT OF SULPHUR COMPOUNDS ON CO2 AND AIR REACTIVITY OF PETROL COKE Yaw Delali Bensah1 and Trygve Foosnaes2 1Department of Materials Science and Engineering, University of Ghana, Accra-Ghana 2Department of Materials Science and Engineering, Norwegian University of Science and Technology, Trondheim-Norway E-Mail: bensahyad@yahoo.co.uk ABSTRACT Three different single-source coke types (SSA, SSB and SSC) were studied for their air and CO2 reactivities using the Hydro method and the observed correlation with reactivity determinant parameter such as elemental composition was noted. The reaction temperatures were 525 oC and 960 oC for air and CO2 reactivities, respectively. Coke sample SSA recorded the highest CO2 reactivity value of 166 mg/g h while coke sample SSB recorded the lowest value of 25 mg/g h. The coke CO2 reactivity showed a moderately strong correlation with the combined effect of Na, Fe and Ca concentrations. The inhibitory effect of sulphur as a catalytic poison on Na was observed with significant downward trend in CO2 reactivity of the investigated cokes. Coke air reactivities showed the expected strong correlation with V. Air reactivity was highest in sample SSC (262 mg/g h) at V concentration of 378 ppm and lowest in sample SSB (39.6 mg/g h) at V concentration of 68 ppm. Sample SSA recorded reactivity value of 129.6 mg/g h at V concentration of 147 ppm. The compound 1-butanethiol was identified by 1-D 1H NMR and 13C NMR, and by 2-D COSY, HSQC and HMBC NMR spectroscopic techniques. It is proposed that 1-butanethiol is one of the possible organosulphur compounds responsible for the reaction with Na forming a stable non-mobile complex partially inhibiting the catalytic effect of Na. Keywords: petrol coke, reactivity, organosulphur compounds. INTRODUCTION consumption represents an area in which significant cost Anode quality has one of the largest variable reductions can be made [7]. impacts on the aluminium smelting cost with the raw In this respect producers of prebaked anodes for materials strongly influencing the cell operation aluminium electrolysis are continually seeking to improve parameters. The goal of many anode manufacturers is to anode quality in order to reduce aluminium production process the materials for optimum properties in order to costs [8]. Quantifying coke reactivity can be used to meet the anode requirements of aluminium smelters [1]. correlate values of anode reactivity data. If reliable However, the susceptibility of petroleum coke to airburn correlations exist, then CO2 and air reactivities may be and CO2 attack (measured and reported as air and CO2 used in predicting anode reactivity [9]. This can be useful reactivities, respectively) are influenced by the feed in selecting raw materials for the manufacture of material used in the petroleum refining and the coke commercial anodes with a potential of high resistance to processing during delayed coking which are determined by reactivities of CO2 and air. two basic parameters; purity and structure [2]. Poor anode On the other hand, during calcining of high quality leads to a decline of cell operation parameters such sulphur coke, compounds of sulphur are driven off and as current efficiency, consumption of energy, anodes and condense in the coke pores and on the surface. These fluorides, metal quality, amount of work involved etc. [3]. compounds become active when making anodes of high Impurities from the crude oil such as metals and sulphur Na concentration. These sulphur compounds are readily tend to be concentrated in the heavy residue and hence in available to Na+ to form sulphide or sulphate compounds the petroleum coke [4]. Reports [5] show that anode during anode baking, partly inhibiting the detrimental carbon can become second to alumina costs as the most effect of Na+ as an oxidation catalyst [10]. It is probable important expenditure in aluminium production. Studies that sulphur acts as a gasification inhibitor in such show that anode costs account for about 20 % of the total reactions since the element can form very stable metal cost of aluminium production, and so reducing anode sulphide compounds and is known to be a potent metal consumption can have an important impact on a smelter’s catalyst poison [11]. It is postulated that the inhibitive profitability [6]. effect of sulphur on the Na catalyzed CO2 gasification of Air and CO2 reactivities have been characterized anodes is caused by the formation of a stable non-mobile to be the determinant of the greater part of the excess Na-S-O complex that prevents Na from catalyzing the carbon consumption per tonne of aluminium produced. It reaction at the active sites [12]. Knowledge of the nature is known that excess consumption arises mainly from of these sulphur compounds can improve the either direct oxidation of the anode by oxygen or oxidation understanding of the mechanism and the inhibition effect of the anode by CO2 which occurs in the pores of the on CO2 reactivity. anode or on the grains of carbon dust floating in the bath In this work air and CO2 reactivities of carbon of the aluminium electrolytic cell. Monitoring the anode samples were studied by the Hydro method. In addition studies were carried out on the extraction and structural 35 VOL. 5, NO. 6, JUNE 2010 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences ©2006-2010 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com identification of the organosulphur compounds from the Ltd, Dublin-Ireland); ethanol, 96 % (Arcus A.S, Oslo - coke samples that react with Na+, forming stable inactive Norway); ethyl acetate, 100 %; n-hexane, 97 % ( all from species inhibiting the catalysis of CO2 gasification BDH Chemicals Ltd, Poole-England); CDCl3, 99.8 % reaction. NMR spectroscopic techniques were used for the (Sigma-Aldrich, Missouri-U.S.A); Al TLC silica gel 60 structural identification of the organosulphur compounds. F254 (Merck KGaA, Darmstadt Germany). Single source The elemental composition of the three coke samples were crude calcined cokes A, B and C (SSA, SSB and SSC) determined by X-ray Fluorescence method whiles the were used for the reactivity tests. number of compounds found in the extracts of the coke samples were determined by thin layer chromatography Experimental procedure for CO2 and air reactivities (TLC). The CO2 and air reactivity tests were carried out using the Hydro method as shown diagrammatically in MATERIALS AND METHODS Figure-1. The sample particle sizes (x) for the reactivity measurements fell in the range 150 < x > 250 µm and Materials these were maintained in all the experiments for the three The chemicals used in this study were nitrogen, coke types. For both CO2 and air reactivities, five parallel 99.999 %; oxygen, 99.999 %; carbon dioxide, 99.95 % (all experiments were run for each coke type (SSA, SSB and from AGA AS, Oslo-Norway); acetone, 99.8 % (Labscan SSC). Figure-1. Schematic representation of the experimental setup. Figure-1 is the schematic representation of the which the gas was switched back to N2. The recorder experimental setup. The experiments were carried out displayed a linear weight change against time. The same under isothermal conditions. The furnace type employed procedure was repeated for air reactivity using air as the was a gold-coated quartz tube-furnace. The lower part of reaction gas at a temperature of 525 oC. the sample holder acted as the control thermocouple. The recorder (personal computer) connected to the balance Experimental procedure for spectroscopic and continuously recorded mass changes for every 10 seconds chromatographic analyses to the nearest of 0.001 g. In the actual experiment coke The extraction of the adsorbed sulphur sample of 1 g was transferred to the sample holder. For the compounds, the coke samples of particle sizes 104 < x > CO2 reactivity experiments the furnace was heated and 250 µm (SSA, SSB and SSC) were weighed into batches kept at 960±1 oC. Simultaneously, the furnace was purged of 100 g each. A given batch was refluxed with acetone with nitrogen (N2) gas at a constant flow rate of 1.5 l/min for 8 hours using a Soxhlet apparatus. until a stable temperature and mass of the sample were For nuclear magnetic resonance (NMR) analysis observed. The N2 flowing to the furnace was switched off the sample extracts obtained above from the three coke and CO2 admitted into the furnace at the same flow rate to samples were evaporated to dryness. The resulting solids start the reaction. The weight of the sample was were dissolved in 1 ml deuterated chloroform (CDCl3). automatically recorded every 10 seconds for 3 hours after The samples were transferred into 5 mm diameter NMR 36 VOL. 5, NO. 6, JUNE 2010 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences ©2006-2010 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com tubes to a height of 5 cm. Spectra for 1H NMR, 13C NMR 200 µm thick, 5×10 cm silica-gel- coated aluminium plate and 2-D NMR were obtained from 300 and 400 MHz and the mobile phase was a 1:1 mixture of ethyl acetate (DPX300 and DPX400 Avanc Bruker) NMR and n-hexane. The separated components of the samples spectrometers. were observed and counted under room and ultraviolet The one dimensional (1-D) proton NMR (1H) and light. The elemental compositions of the petrol cokes were 13C NMR signals (Figures 3 and 4) obtained were difficult determined using XRF. to interpret due to their complex spectra resulting from the mixture of compounds where the spectra have many RESULTS coincident resonances (overlap). These difficulties were In order to determine the reactivities the exposure circumvented by employing two dimensional (2-D) NMR times to CO2 and air were noted. Plots of weight loss procedures which were used in the interpretation. against time were made. The reactivities (R) were Information obtained from 2-D data was used to derive computed from the slope of the linear regression plot from firm conclusions about connectivities between 1H-1H, 1H- the last half (1.5 hours) because during this period the 1 3C, 13C-13C subunits. The 2-D techniques used were system is stabilized for both CO2 and air reactivities for all Correlation Spectroscopy (1H-1H COSY), Heteronuclear the five parallel experiments for samples SSA, SSB and Multiple Bond Correlation (HMBC) and Heteronuclear SSC. Using the graphical plot of experiment one of sample Single Quantum Correlation (HSQC) [13]. For the thin SSA as a representative plot shown in Figure-2, the layer chromatographic analyses, the stationary phase was reactivities were computed from equations 1, 2 and 3. Figure-2. Weight loss as a function of time for experiment SSA1. Where ∆W, a, mi, t and b are the weight loss (g), a constant (gradient), the initial mass of sample (g), time (min) and intercept respectively. 37 VOL. 5, NO. 6, JUNE 2010 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences ©2006-2010 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com Table-1. Reactivity values determined for sample SSA. R with CO R with air Expts 2 mg/gh mg/cm2h mg/gh mg/cm2h SSA1 174 45.8 132 34.7 SSA2 162 42.6 126 33.1 SSA3 168 44.2 132 34.7 SSA4 180 47.3 126 33.1 SSA5 150 39.4 132 34.7 Mean 166.8 43.9 129.6 34.1 Std dev 10.3 2.7 2.9 0.8 Table-2. Reactivity values determined for sample SSB. R with CO2 R with air Expts mg/gh mg/cm2h mg/gh mg/cm2h SSB1 24 6.3 42 11 SSB2 24 6.3 42 11 SSB3 30 7.9 36 9.5 SSB4 30 7.9 42 11 SSB5 18 4.7 36 9.5 Mean 25.2 6.6 39.6 10.4 Std dev 4.5 1.2 2.9 0.4 Table-3. Reactivity values determined for sample SSC. R with CO2 R with air Expts mg/gh mg/cm2h mg/gh mg/cm2h SSC1 84 22.1 264 69.4 SSC2 84 22.1 256 67.3 SSC3 72 18.9 276 72.6 SSC4 96 25.2 252 66.3 SSC5 84 22.1 264 69.4 Mean 84 22.1 262.4 69 Std dev 7.6 2 8.2 2.2 Table-4. The elemental composition of samples. Element Na Mg Al Si S (%) K Ca Ti V Mn Fe Ni Zn SSA (ppm) 55 12 54 63 1.2 3 110 3 147 4 105 82 2 SSB (ppm) 28 4 61 105 1 6 42 3 68 4 189 44 4 SSC (ppm) 43 8 24 23 2.6 2 27 1 378 1 378 170 2 38 VOL. 5, NO. 6, JUNE 2010 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences ©2006-2010 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com Figure-3. Proton (1H) NMR spectrum of sample SSA in CDCl3 run at 300 MHz. Figure-4. Carbon (13C) NMR spectrum of sample SSA in CDCl3 run at 300 MHz. As obtained in Figure-5, the COSY spectrum diagonal and the projection on each axis show the one shows a correlation between directly coupled protons, i.e. dimensional spectrum. The off-diagonal peaks indicate the interactions between protons on adjacent carbons. The presence of coupling between pairs of protons. 39 VOL. 5, NO. 6, JUNE 2010 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences ©2006-2010 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com Figure-5. 1H-1H correlation spectroscopy (COSY) spectrum of sample SSA in CDCl3 run at 400 MHz. The HSQC spectrum in Figure-6 which is also summarizes the strong proton-carbon coupling from the proton detected gives strong proton-carbon coupling i.e. HSQC spectrum of Figure-6. the protons directly bonded to the carbon. Table-5 Figure-6. Heteronuclear single quantum correlation (HSQC) spectrum of sample SSA in CDCl3 (100 MHz for 1H, 400 MHz for 13C). 40 VOL. 5, NO. 6, JUNE 2010 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences ©2006-2010 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com Table-5. HSQC correlation data for sample SSAA showing direct proton-carbon correlations as obtained from Figure-6. 1H (ppm) 2 2 5.4 8.1 7.6 1.6 1.6 2.3 4.1 2.65 13C (ppm) 21 26 130 130 130 26 28 32 64 54 1H (ppm) 2.15 4.9 0.85 0.85 0.85 1.1 1.2 0.7 1.35 1.25 13C (ppm) 31 67 14.1 19.5 22.7 37 37 11.5 17 14 Information obtained from HMBC spectrum in proton-carbon coupling indicates that the proton is two or Figure-7 below, capitalizes on weak proton-carbon three bonds away from the carbon. The deduced proton- coupling and suppresses one-bond correlations. A weak carbon correlations are summarized in Table-6. Figure-7. The heteronuclear multiple bond correlation (HMBC) spectrum of sample SSA in CDCl3 (100 MHz for 1H, 400 MHz for 13C). Table-6. HMBC correlations data for sample SSA showing proton-carbon multiple bond correlations as obtained from Figure-7. 1H (ppm) 4.1 2.3 8.1 8.1 4.1 4.1 2 2 1.6 1.6 13C (ppm) 174 174 134 165 26 28 170 32 26 28 1H (ppm) 2.3 0.7 7.9 0.7 0.75 1.05 1.3 1.25 1.25 1.25 13C (ppm) 28 23 130 20 14.1 14.1 14.1 69 54 22 1H (ppm) 3.9 2.65 8.3 2.65 2.5 2.8 1.9 2.2 2.7 0.75 13C (ppm) 64 29.5 130 70 54 54 21 21 20 23 1H (ppm) 1.05 1.25 0.9 0.9 0.9 2.15 1.45 1.15 2.3 4.2 13C (ppm) 23 23 23 28 33 35 23 23 25 64 DISCUSSIONS most important influence are iron [18], sodium [19] and The TLC tests identified 9 different compounds calcium [20]. The CO2 reactivity was high in sample SSA present in the extracts. The TLC results support proton (166 mg/g h) due to high levels of Fe, Na, Ca and low in NMR spectrum which shows that the samples are mixtures sample SSB (25 mg/g h) due to low concentrations of Na of different compounds with overlapping and complex and Ca. The low reactivity value of sample SSB was due peaks. to low elemental composition. The coke CO2 reactivity For air reactivity the elements of most important showed a moderate correlation with Na, Fe and Ca as influence are vanadium, nickel [15], sodium [16], lead and shown in Figure-8. copper [17] whereas for CO2 reactivity the elements of 41 VOL. 5, NO. 6, JUNE 2010 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences ©2006-2010 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com 50 40 30 20 10 0 125 175 225 275 325 Σ(Na, Fe, Ca) ppm Figure-8. Correlation between CO2 reactivity and Σ (Na, Fe, Ca) content for investigated cokes. Air reactivity was high in sample SSC (262 mg/g largely to the high concentration of V (378 ppm), and h) and low in sample SSB (39.6 mg/g h). This matched the partially by the concentrations of Na (43 ppm), and Ni expected strong correlation with the Vanadium (V) (170 ppm). Furthermore, the low air reactivity value for composition of the samples as shown in Figure-9. Sample sample SSB was due to the low levels of V (68 ppm), Na SSC recorded the expected high air reactivity value due (28 ppm) and Ni (44 ppm). 80 70 60 50 40 30 20 10 0 0 50 100 150 200 250 300 350 400 450 V (ppm) Figure-9. Correlation between air reactivity and V content for investigated cokes. From Figure-3, the aliphatic sulphur containing Figure-6. Furthermore, the δ1.6 ppm exhibits a two bond functional group of interest may occur at a chemical shift correlation to the δ26 ppm carbon as in Figure-5 and (δ) range of 1.2-1.8 ppm. From Figure-4 the carbon tabulated in Table-5. This indicates that the suspected containing the expected functional group of interest falls organosulphur compound is a thiol with the sulphur atom within the range of δ5-45 ppm. In Figure-5 the protons at sandwiched between carbon at δ26 ppm and a sulphhydryl δ1.6 ppm are coupled to protons at δ2.3 ppm and δ4.1 proton of δ1.6. The proton at δ2.0 ppm is also bonded to ppm. Sulphhydryl protons can usually exchange at a low carbon at δ26.0 ppm. Furthermore, the δ1.6 ppm exhibits a rate so that at room temperature they are coupled to two bond correlation to the δ26 ppm carbon as shown in protons on adjacent carbon atoms [13]. This phenomenon the Figure-5. From Figure-5, the proton δ4.1 ppm is next was observed as the proton chemical shift of 1.6 ppm was to the proton δ1.6 ppm which in turn is adjacent to proton directly attached to carbon at δ26 ppm as is evident in δ2.3 ppm. The HSQC data in Figure-6 shows that proton 42 Air reactivity (mg/cm2h) CO2 reactivity (mg/cm2h) VOL. 5, NO. 6, JUNE 2010 ISSN 1819-6608 ARPN Journal of Engineering and Applied Sciences ©2006-2010 Asian Research Publishing Network (ARPN). All rights reserved. www.arpnjournals.com δ4.1 is directly bonded to carbon at δ64 ppm, proton δ1.6 [6] Baber A. M., Proulx L. A. 1994. Anode consumption ppm is directly bonded to carbon at δ28 ppm, and proton study. Light Metals. p. 677. δ2.3 ppm is directly bonded to carbon at δ32 ppm. Furthermore, Figure-7 and its corresponding Table-6, [7] Billehaug K., Øye H. A. 1981. ALUMINIUM 57. p. shows that carbon at δ32 ppm is joined to carbon at δ26 146 and p. 228. ppm by the formation of a three bond correlation between proton δ1.6 ppm and carbon δ28 ppm. Proton δ2.3 ppm is [8] Samanos B., Dreyer C. 2001. Impact of Coke linked to carbon at δ28 ppm by two bond correlation in Calcination level and Anode Baking Temperature on Table-6. A two bond correlation is also formed between Anode Properties. Light Metals. p. 681. the proton δ4.1 and carbon at δ28 ppm. This identifies the organosulphur as 1-butanediol, n-C4H10S. [9] Rolle G. J. and Czikall A.R. 2001. Use of Coke Air Reactivity Testing for Predicting Anode Air 4.1 1.6 2.3 2.0 Reactivity. Light Metals. p. 675. H H H H _⏐_⏐_ ⏐_ ⏐_ _ [10] Foosnæs T. 2007. Private communication. 15 January. 4.1H C64 C C C S ⏐ ⏐28 ⏐32 ⏐26 H 1.6 [11] Maxted E. B. 1951. The poisoning of metallic H H H H catalysts. Adv Catal. 3: 129. 4.1 1.6 2.3 1.6 [12] Hume S. M., Fischer W. K., Perruchoud R. C., CONCLUSIONS Metson J. B., Baker R. T. K. 1993. Influence of Among the determinant parameters that affect petroleum coke sulphur content on the sodium coke reactivity, the elemental composition of the coke sensitivity of carbon anodes. Light Metals. p. 535. remains the single most important parameter. The high V content in the cokes had a significant impact on air [13] Silverstein R. M. and Webster F. X. 1998. reactivity and showed a strong correlation. Air reactivity Spectrometric Identification of Organic Compounds. th was highest in sample SSC (262 mg/g h) at a V 6 Ed. NY: Wiley interscience publication. p. 2, 14, concentration of 378 ppm and lowest in sample SSB (39.6 69, 144, 168, 232, 255-264. mg/g h) at a V concentration of 68 ppm. Coke sample SSA recorded the highest CO2 reactivity value of 166 mg/g h [14] Crews P., Jaspars M., Rodrigue J. 1998. Organic st while coke sample SSB recorded the lowest value of 25 structure analysis. 1 Ed. NY: Oxford University mg/g h. The coke CO2 reactivity further showed a Press. p. 7, 23, 28-32, 230-233. moderately strong correlation with the combined effects of Na, Fe and Ca concentrations. The compound 1- [15] Reis T. 1977. Chem. Tech. 7(6): 336. butanethiol was identified as a possible organosulphur compound responsible for inhibiting the catalytic effect of [16] Lazarev V. D., Yanko E. A., Zakharov V. V. 1978. Na on CO reactivity. Sov. J. Nonferrous Metals. 19(2): 38. 2 REFERENCES [17] Sekhar J. A. and Liu J. 1999. Preferential Oxidation Processes of Carbons used in the Hall-Héroult Cell. [1] Meier M. W. 2006. The impact of raw material quality Light Metals. p. 531. and anode manufacturing parameters on the behaviour in electrolysis. 25th International Course on Process [18] Kolodin E. A., Nikitin V. 1982. Ya, Tsvet. Met. (8): Metallurgy of Aluminium, Trondheim-Norway, May 44. 29 - June 02. 1: 133-134. [19] Wilkening. S. 1994. Why the sulfate reactivity test. [2] Meyers R. A. 1986. Handbook of Petroleum Refining Light Metals. p. 643. Processes. McGraw Hill. [20] Rey Boero J. F. 1981. Studies on anode reactivity to [3] Perruchoud R. C. 2001. Anode properties cover oxidant gas. Light Metals. p. 441. materials and cell operation. Light Metals. p. 695. [21] Chmelar J. 2006. Size reduction and specification of [4] Fischer W. K. and Perruchoud R.C. 1985. Influence of granular petrol coke with respect to chemical and Coke Calcining Parameters on Petroleum Coke physical properties. PhD Thesis. NTNU. p. 15. Quality. Light Metals, AIME, New York. p. 811. [5] Keller F., Fischer W. K. Consumption of anode carbon during aluminium electrolysis. Light Metals. p. 729. 43