The Influence of Dimethyl Disulfide on Naphtha Steam Cracking

Natália OlahováStamatis A. SarrisMarie-Françoise ReyniersGuy B. MarinKevin .... E. Mahmoudi , A. Hafizi , M.R. Rahimpour , A. Bolhasani , A. Sharia...
0 downloads 0 Views 146KB Size
Ind. Eng. Chem. Res. 2001, 40, 4353-4362

4353

The Influence of Dimethyl Disulfide on Naphtha Steam Cracking Inge Dhuyvetter, Marie-Franc¸ oise Reyniers, Gilbert F. Froment, and Guy B. Marin* Laboratorium voor Petrochemische Techniek, Universiteit Gent, Krijgslaan 281 S5, B-9000 Gent, Belgium

Dominique Viennet ATOFINA, Centre de Recherches et De´ veloppement de l’Est, BP 61005, 57501 Saint-Avold Cedex, France

A pilot unit for steam cracking equipped with a transfer line heat exchanger (TLE) that allows for the study of coke deposition in both the reactor and the TLE is presented. The reactor and TLE are made of Incoloy 800HT. The duration of a coking run typically amounts to 32.4 ks. The influence of different dimethyl disulfide (DMDS) addition procedures, i.e., continuous addition, presulfidation and presulfidation followed by continuous addition, on CO production and on coke deposition in the reactor and in the TLE during naphtha cracking is investigated. Presulfidation reduces CO production. However, to obtain a low and stable CO production, continuous addition of sulfur is required. The influence of sulfur addition on coke formation in the reactor can strongly differ from its influence on coke formation in the TLE. In the reactor, as well as in the TLE, the observed influence of sulfur addition is complex and strongly depends on the technique used. The optimal operating conditions for reducing CO production and minimizing coke formation consist of presulfidation followed by continuous dosing. 1. Introduction Thermal cracking of hydrocarbon feedstocks ranging from ethane to gasoil is the main source of olefins and aromatics, which are important feedstocks in the petrochemical industry. Cracking is carried out in tubular reactors constructed of heat-resistant Fe-Ni-Cr alloys at extreme conditions of temperature (800-950 °C). The gases leaving the reactor coil are rapidly cooled in a transfer line exchanger (TLE) in which steam is generated. Thus, cooling is a major factor contributing to the economics of an ethylene plant because it prevents further reactions of valuable products such as ethylene and butadiene and recovers heat from the effluent gases. TLEs are generally constructed in low alloyed steels such as 15Mo3. An inherent problem associated with the construction materials used in ethylene plants is their tendency to promote the formation of carbonaceous materials that accumulate in the reactor coil as well as in the TLE. The accumulation of coke during the large-scale thermal cracking of hydrocarbons leads to decreased heat transfer, a reduction of the tube cross section, and an increased pressure drop. The loss of the furnace availability due to decoking, the decrease of the olefin selectivity and the energy losses associated with the accumulation of coke on the reactor wall have important negative consequences for the economics of the cracking process. Moreover, carburization of the tubes can occur, leading to materials damage. Coke deposition in the TLE increases the exit temperature of the process gas and the pressure drop over the TLE and interferes with the production of high-pressure steam. Carbon monoxide, produced during steam cracking, acts as a poison for the catalyst, usually Pd supported on alumina, used in the downstream hydrogenation of * To whom correspondence should be sent. E-mail address: [email protected]. Fax: 0032/9/2644999. Tel.: 0032/9/ 2644516

acetylene. To obtain on-specification ethylene at the outlet of the hydrogenation reactor in the presence of high CO concentrations, high operating temperatures are required.1 A sudden reduction in the CO content can cause an uncontrolled increase in the hydrogenation reactor temperature. In industrial practice, additives are frequently used to control CO production.2 The most widely used group of additives in commercial ethylene plants is based on sulfur components. Sulfur components occur naturally in some liquid feedstocks and are added in gas cracking. In addition to reducing the CO production, sulfur addition is believed to minimize the overall coking rate3 by suppressing the catalytic activity of the metal wall.4 Three different techniques are used for the addition of sulfur additives: presulfidation, continuous addition, and presulfidation followed by continuous addition. However, detailed operating procedures are considered proprietary and are therefore largely unavailable in the open literature. Also, information on the influence of sulfur on coke formation in industrial units is scarce or nonexistent. Direct data on coke formation mainly stem from observations made after a catastrophic shutdown. Usually, either the increase in pressure drop measured over the reactor and the TLE or the CO content in the effluent is taken as an indicator for coke formation. The observed rapid increase in pressure drop when sulfur addition is interrupted is taken as an indication of increased coking in the absence of sulfur. Clearly, this type of measurement raises the question as to whether the increased coke deposition occurs in the coil and/or in the TLE. Moreover, the reliability of these types of measurements as an indicator for coke formation is questionable. As discussed by Reyniers and Froment,5 the reduction of CO production in the presence of sulfur cannot be taken as an indication of reduced coke formation. The observations of Reed6 indicate that, in the absence of sulfur, the pressure drop can increase

10.1021/ie001131b CCC: $20.00 © 2001 American Chemical Society Published on Web 08/31/2001

4354

Ind. Eng. Chem. Res., Vol. 40, No. 20, 2001

Figure 1. Schematic representation of the reactor and the cooling section.

rapidly as a result of a very rapid and localized coke deposition in the transfer line/TLE area, notwithstanding the fact that, overall, substantially less coke is produced than in the presence of sulfur. Most studies on the influence of sulfur on coke formation reported in the open literature pertain to the cracking coil and to observations at temperatures between 700 and 1000 °C.5,7-12 The observations are contradictory. However, these studies seem to indicate that not only the amount and type of sulfur compound added but also the technique used to add the sulfur compound, i.e., presulfidation, continuous addition, or a combination of the two, can play an important role in determining the rate of coke deposition. Far less attention has been devoted to the influence of sulfur on coke formation under TLE conditions. To our knowledge, the only data available are those reported by Reed,6 which were obtained in bench-scale experiments. In this study, the influence of sulfur on CO production and coke formation in both the reactor and the transfer line exchanger during the steam cracking of naphtha feedstocks is reported. The experimental data were obtained in a pilot plant unit with a TLE section allowing for the separate determination of coke deposition in the reactor and in the TLE. In addition to the influence of sulfur naturally occurring in naphtha, the influence of added dimethyl disulfide (DMDS) was investigated because it is frequently used in industrial practice. The following techniques of DMDS addition were investigated: continuous addition, presulfidation, and presulfidation followed by continuous addition. 2. Experimental Section 2.1. Pilot Plant. All data reported in this study were obtained in the pilot plant for the thermal cracking of

hydrocarbons at the Laboratorium voor Petrochemische Techniek of Ghent University. To study coke deposition in the TLE, the cooling section of the pilot unit was modified. Because the unit, the analytical equipment, and the calibration procedure have been described elsewhere,13-15 only a brief description with emphasis on the TLE section will be given here. The furnace, built of silica/alumina brick (Li23), is 4 m long, 0.7 m wide, and 2.6 m high. It is fired by means of 90 premixed gas burners, mounted with automatic fire checks and arranged on the side walls in such a way as to provide a uniform distribution of heat. The fuel supply system comprises a combustion controller for the regulation of the fuel-to-air ratio and the usual safety devices. The furnace is divided into seven separate cells that can be fired independently to set any type of temperature profile. Twenty thermocouples and five manometers are located along the reactor coil to measure the temperature and pressure of the reacting gas. The reaction section of the tube is 12.8 m long and has an internal diameter of 9 mm. These dimensions where chosen to achieve turbulent flow conditions in the coil with reasonable feed flow rates. The cooling section consists of two heat exchangers, TLE1 and TLE2. TLE1, made of Incoloy 800HT, is designed to achieve turbulent flow conditions with effluent flow rates typical for the pilot unit and is used to study coke deposition under TLE conditions. It consists of two concentric tubes: the reactor effluent flows through the inner tube, while air, providing cooling of the effluent, flows cocurrently through the outer tube. A schematic representation of the reactor and the cooling section is given in Figure 1; the dimensions of TLE1 are given in Table 1. Both air and

Ind. Eng. Chem. Res., Vol. 40, No. 20, 2001 4355 Table 1. Dimensions of TLE1

thermowell process gas tube air cooling tube

Table 2. Feedstock Characteristics

inner diameter (mm)

outer diameter (mm)

length (mm)

4.31 22.10 35.00

6.35 25.40 39.00

1550 1550

the process gas enter at the top of TLE1. Cocurrent flow of the two streams was chosen because it provides a more uniform wall temperature profile along the TLE as compared to countercurrent flow. By adjusting the air flow rate, the temperature profile of the process gas in TLE1 can be regulated. TLE1 can also be heated to 900 °C for decoking with air/steam. In TLE2, which is a concentric tube heat exchanger, the process gas is further cooled to 150 °C by means of a cooling oil. A coking run consists of three stages: (1) The first stage involves cracking and coke deposition in the reactor and in TLE1. During this stage, the reactor effluent flows successively through TLE1 and TLE2 and coke is deposited in the coil and in TLE1. At the outlet of TLE1, the injection of nitrogen provides an internal standard. Sampling for C5+ analysis occurs downstream of TLE1, and that for C4- analysis occurs downstream of TLE2. (2) In the second stage, decoking of the reactor with a steam/air mixture occurs. Decoking of the reactor coil is performed with a steam/air mixture. During this stage, N2 is sent countercurrently through TLE1, and a connection with the COx analyzers is established. During decoking of the reactor coil, the CO and the CO2 contents in the effluent are monitored continuously with IR meters. A vortex gas meter measures the volumetric gas flow rate. These data are used to calculate the total amount of carbon deposited in the reactor. It should be mentioned that the rates of coke deposition reported in this study represent values averaged over the length of the reactor and over the duration of the run. In a tubular reactor, coke is deposited according to a profile that depends on the product distribution and on the temperature at each point in the reactor. Thus, in a tubular reactor only average rates of coking can be obtained. (3) The last stage involves decoking of TLE1 with a steam/air mixture. Because the reactor has been decoked previously, the steam/air mixture is heated in the reactor before entering TLE1. The IR meters determine the CO and the CO2 contents in the effluent while the deposited coke is being burned off. A vortex gas meter measures the volumetric gas flow rate. These data are used to calculate the total amount of carbon deposited in TLE1. As for the reactor, the rates of coke deposition in TLE1 represent values averaged over the length of TLE1 and over the duration of the run. 2.2 Analysis Section of the Pilot Plant. The C4sample is simultaneously analyzed on two gas chromatographic (GC) systems. Only hydrogen is not detected twice. The use of two different units for the same analysis improves the reliability of the results. The first system consists of a Carle CGC 530 automated refinery gas analyzer. The carrier gas is helium. Nitrogen, carbon monoxide, carbon dioxide and hydrocarbons up to C2 are detected by a thermal conductivity detector (TCD), C3 and C4 by a FID. This chromatograph is also equipped with a hydrogen transfer system, a palladium tube that allows for the migration of hydrogen from helium into nitrogen carrier gas. In nitrogen, the hydrogen is easily detected by a TCD.

naphtha I

naphtha II

naphtha III

molar volume molar mass (g/mol) density (kg/m3)

128.1 97.8 763

136.1 107.4 789

144.3 111.3 771

paraffins (wt %) isoparaffins (wt %) aromatics (wt %) naphthenes (wt %) olefins (wt %)

17.1 23.7 12.2 46.5 0.3

12.2 19.2 15.7 52.8 0

20.3 27.9 14.5 36.9 0

ppm of S

20

0

0

(cm3/mol)

The second system consists of an Interscience Fisons GC 8340 instrument with a TCD for the analysis of nitrogen, carbon monoxide, carbon dioxide, and hydrocarbons up to C2 and a Packard 437A chromatograph with FID for the analysis of all hydrocarbons up to C4. The C4- analysis takes about 30 min. Because the C5+ analysis lasts for more than 70 min, two C5+ chromatographs (HP 5890 Series II) are used in alternate operation. These instruments are used for the analysis of all hydrocarbons above C4. Peak identification and integration is performed by a commercial integration package (XChrom of Labsystems). Calculations are based on the absolute flow rates of the effluent components. This is made possible by the injection of a precisely known nitrogen flow. From the peak areas of the Carle TCD, the experimentally determined calibration factors of this instrument, and the known amount of nitrogen, the flows of hydrogen, methane, COx, and C2 hydrocarbons are calculated. Using the methane flow thus calculated, the flows of the other components can be calculated on the other instruments. Because both nitrogen and methane are also detected on the Interscience Fisons GC 8340 instrument, a check for the calculated methane flow on the Carle TCD is available. With these data, a product distribution in terms of weight percentages can be determined. Because the feed flow rate is known, yields and a material balance can also be calculated. 2.3. Operating Conditions. 2.3.1. Reactor and TLE Characteristics. The reactor, TLE1, and TLE2 are made of Incoloy 800HT (Ni, 30-35; Cr, 19-23; and Fe, >39.5 wt %). In the radiant coil, pyrolysis and coke deposition are considered to occur only in cells where T > 600 °C. For the temperature profiles used in this study, the reactor surface area available for coke deposition amounts to 0.258 or 0.340 m2. The TLE1 surface area amounts to 0.143 m2. 2.3.2. Feed. Three different feedstocks were used. The main characteristics of the feedstocks are presented in Table 2. Feed I, denoted naphtha I, is a naphtha containing 20 ppm of sulfur; C6, C7, C8, and C9 hydrocarbons are the main constituents and total up to approximately 87 wt %. Feeds II and III, denoted naphthas II and III, are sulfur-free naphthas; C7, C8, C9, and C10 hydrocarbons amount to ∼88 wt % for naphtha II and to ∼92 wt % for naphtha III. 2.3.3. Pretreatment Conditions. 2.3.3.1. Steam Pretreatment Procedure. Because coke deposition strongly depends on the state and composition of the surface, a standardized pretreatment procedure was used prior to all experiments. The reactor is heated under a steam flow rate of 1.1 g/s. After 900 s, the temperature profile in the reactor and in TLE1, as specified by the cracking conditions (see

4356

Ind. Eng. Chem. Res., Vol. 40, No. 20, 2001

Table 3. Operating Conditions during Presulfidation with DMDS test

4

5

6

7

8

9

steam flow rate (g/s) DMDS flow rate (mg/s) S (mg of S/s) duration (s)

2 0.18 0.12 1800

2 0.7 0.48 1800

2 1.39 0.95 1800

2 0.7 0.48 1800

2 1.39 0.95 1800

2 2.78 1.89 1800

T profile of the reactor outlet cell 3 (K) outlet cell 4 (K) outlet cell 5 (K) outlet cell 6 (K) COT (K)

772 871 1026 1085 1138

772 871 1026 1085 1138

772 871 1026 1085 1138

772 871 1026 1085 1138

772 871 1026 1085 1138

772 871 1026 1085 1138

T profile of TLE1 TL (K) inlet TLE (K) TLE T2 (K) TLE T3 (K) outlet TLE (K) Tmean (K)

1104 985 761 704 643 773

1104 985 761 704 643 773

1104 985 761 704 643 773

1104 985 761 704 643 773

1104 985 761 704 643 773

1104 985 761 704 643 773

Table 4. Cracking Conditions 1

2

3

feed HC flow rate (g/s) steam flow rate (g/s) dilution (kg/kg) COP (bar) residence time (ms)

test

naphtha I 1.33 0.67 0.5 1.7 285

naphtha II 1.33 0.67 0.5 1.7 287

naphtha III 1.33 0.67 0.5 1.7 200

T profile of the reactor outlet cell 3 (K) outlet cell 4 (K) outlet cell 5 (K) outlet cell 6 (K) COT (K)

871 948 1008 1068 1118

871 948 1008 1068 1118

772 871 1026 1085 1138

T profile of TLE1 TL (K) inlet TLE (K) TLE T2 (K) TLE T3 (K) outlet TLE (K) Tmean (K)

936 757 692 643 757

1074 933 763 702 643 760

1104 985 761 704 643 773

duration (ks) reactor surface area (m2) TLE1 surface area (m2)

21.6 0.340 0.143

32.4 0.340 0.143

32.4 0.258 0.143

section 2.3.4), is set, and the steam flow rate is set at 2 g/s. After 5.4 ks, the desired temperature profiles are reached, and then the steam flow rate is set at 0.67 g/s. The hydrocarbon flow rate is set at 1.33 g/s and admitted to the reactor. 2.3.3.2. Presulfidation with DMDS. The reactor is heated under a steam flow rate of 1.1 g/s. After 900 s, the temperature profile in the reactor and in TLE1, as specified by the cracking conditions (see section 2.3.4), is set, and the steam flow rate is set at 2 g/s. After 5.4 ks, the desired temperature profiles are reached. At this point DMDS is introduced at the inlet of the reactor by means of a ISCO 500D syringe pump. The detailed presulfidation conditions used in this study are specified in Table 3. After 1.8 ks, the steam flow rate is set at 0.67 g/s. The hydrocarbon flow rate is set at 1.33 g/s and admitted to the reactor. 2.3.4. Cracking Conditions. The detailed cracking conditions used in this study are given in Table 4. During coking tests with continuous addition of DMDS, the additive is introduced at the inlet of the reactor by means of a ISCO 500D syringe pump. 2.3.5. Decoking Conditions. 2.3.5.1. Decoking of the reactor coil. Decoking of the reactor coil is performed with a steam/air mixture. During this stage, N2 (0.08

Nl/s) is fed countercurrently through TLE1, and a connection with the COx meters is established. During decoking of the reactor coil, the conditions specified in Table 5a are applied. At the start of the decoking procedure, the reactor is heated to 800 °C under a nitrogen flow, and then steam (0.28 g/s) is introduced to the reactor. After 180 s, the nitrogen flow is stopped, and air (0.23 Nl/s) is admitted to the reactor. Once most of the coke is removed, the temperature is increased to 900 °C. When practically all the coke is gasified, the steam flow is stopped, and decoking occurs in air only. The standard decoking time is 6 ks. 2.3.5.2. Decoking of TLE1. The decoking of TLE1 is performed with a steam/air mixture. During decoking of TLE1, the conditions specified in Table 5b are applied. At the start of the decoking procedure, the coil outlet temperature is kept at 900 °C, and TLE1 is heated under a nitrogen flow. Then, steam (0.35 g/s) is fed to TLE1 via the reactor coil. After 600 s, the nitrogen flow is stopped, and air is admitted to TLE1 via the reactor coil. Once most of the coke is gasified, the temperature in TLE1 is increased. When the CO2 content in the effluent has fallen below 0.1 vol %, the steam flow is stopped, and decoking occurs in air only. Typically, the decoking time is 6 ks. 3. Results and Discussion 3.1. Influence of Sulfur on Coke Formation and CO Production in Naphtha Cracking. To evaluate the influence of sulfur compounds that occur naturally in liquid feedstocks on CO production and on the coking behavior in the coil and in the TLE, a naphtha containing 20 ppmw of S (naphtha I) and a sulfur free naphtha (naphtha II) were cracked under identical conditions (see Table 4, tests 1 and 2). The effluent composition and the data on coke formation are presented in Table 6 (tests 1 and 2). It should be mentioned that, although the duration of the experiments typically amounts to 32.4 ks, the stage of catalytic coke formation was estimated, based on experiments with durations varying from 7.2 to 43.2 ks, to last approximately 10.8 ks. Thus, the observations on the influence of sulfur are not restricted to catalytic coking only. Clearly, because the two effluents have different compositions, a direct comparison cannot be made. However, some general trends on the relation between CO production and coke formation can be seen. Figure 2 presents the CO content in the effluent as a function of run length. During cracking of naphtha II, the CO concentration increases with time to reach a maximum and then decreases to a steady-state value i.e., the asymptotic value. From a mechanistic point of view, two processes are important in describing the steam/coke interaction: the direct gasification of coke with steam and the metal-catalyzed removal of carbonaceous deposits by steam reforming. It is generally accepted that carbon adatoms and CHx groups on the surface are intermediates in the steam reforming of hydrocarbons. The high initial CO production is associated with the catalytic influence of the wall. As the metal particles of the wall become covered with coke, their catalytic activity decreases. However, at the high temperatures prevailing in the coil, the coke is certainly not impervious to the effects of the wall, so that the influence of the metal can still be felt in the asymptotic stage. Reyniers and Froment5 reported a remanent influence of the wall material on the steady-state CO production

Ind. Eng. Chem. Res., Vol. 40, No. 20, 2001 4357 Table 5. Decoking Conditions (a) Reactor Fnitrogen in TLE1 (Nl/s)

Fsteam (g/s)

Fnitrogen (Nl/s)

0.08 0.08 0.08 0.08 0.08

0 0.28 0.28 0.28 0

0.23 0.23 0 0 0

start after 180 s CO2 < 1 vol % CO2 < 0.1 vol %

Fair (Nl/s) 0 0 0.23 0.23 0.23

Tout,cell3 (K)

Tout,cell4 (K)

Tout,cell5 (K)

923 923 923 923 923

1073 1073 1073 1173 1173

1073 1073 1073 1173 1173

Tout,cell6 (K)

Tout,cell7 (K)

1073 1073 1073 1173 1173

1073 1073 1073 1173 1173

(b) TLE1

start after 180 s CO2 < 1 vol % CO2 < 0.1 vol %

Fsteam (g/s)

Fnitrogen (Nl/s)

Fair (Nl/s)

0 0.35 0.35 0.35 0

0.23 0.23 0 0 0

0 0 0.23 0.23 0.23

COT (K) 1173 1173 1173 1173 1173

TLE1 inlet (K)

TLE1 T2 (K)

TLE1 T3 (K)

TLE1 outlet (K)

>1023 >1023 >1023 1073 1073

>1073 >1073 >1073 1173 1173

>1073 >1073 >1073 1148 1148

>973 >973 >973 1023 1023

Table 6. Effluent Composition and Data on Coke Formation test feed HC flow rate (g/s) steam flow rate (g/s) dilution (kg/kg) COT (K) COP (bar) residence time (ms)

1

2

3

naphtha I 1.33 0.67 0.5 1118 1.7 285

naphtha II 1.33 0.67 0.5 1118 1.7 287

naphtha III 1.33 0.67 0.5 1138 1.7 200

hydrogen CO CO2 methane ethylene propylene butadiene benzene toluene xylene C4- (wt %) C5+ up to benzene (wt %) C5+ above benzene (wt %) pyrolysis gasolinea (wt %) pyrolysis gasoilb (wt %) propylene/ethylene

yields (wt %) 0.97 0.01 0.02 11.07 21.16 13.49 6.09 6.93 5.67 3.21 62.53 15.27 15.84 30.35 7.12 0.64

0.87 0.09 0.01 10.97 20.28 11.84 6.32 7.57 6.77 4.16 58.62 14.78 21.02 34.71 6.67 0.58

0.99 0.26 0.04 13.4 25.86 12.14 4.98 8.25 5.85 2.67 65.56 12.69 21.75 28.66 5.78 0.47

mass (g) Rc (mg m-2 s-1) yield (%)

coke reactor 7.41 3.27 1.01 0.3 0.026 0.008

4.74 0.57 0.011

mass (g) Rc (mg m-2 s-1) yield (%)

coke TLE1 3.16 1.02 0.011

1.13 0.24 0.003

1.8 0.39 0.004

a Pyrolysis gasoline ) C + hydrocarbons with boiling points < 5 bp naphthalene (218 °C). b Pyrolysis gasoil ) C5+ hydrocarbons with boiling points g bp naphthalene (218 °C).

during cracking of n-hexane, clearly indicating that steam reforming of hydrocarbons is mainly responsible for CO formation. In the presence of sulfur, however, the CO production is greatly reduced and far more stable as a function of run length. In particular, the maximum at the beginning of the run is completely suppressed by the presence of sulfur. It is well-known that sulfur greatly influences the steam reforming reactions of hydrocarbons.16 Adsorption of sulfur components from the gas phase onto the metal particles proceeds more readily than the adsorption of either hydrocarbons or water. This results

Figure 2. CO content in the effluent as a function of run length during cracking of a sulfur-containing naphtha (Table 5, test 1) and a sulfur-free naphtha (Table 5, test 2).

in a poisoning of the surface and, hence, in a lowering of the catalytic production of CO.17 As can be seen in Table 6, coke deposition both in the reactor coil and in TLE1 is far more pronounced in the presence of sulfur. Our observations indicate that sulfur components present in naphtha control CO production very well. However, there is no direct relation between the amount of CO produced and the amount of coke formed. Thus, monitoring of the CO concentration in the effluent is not a reliable indicator for coke deposition. 3.2. Influence of DMDS on Coke Formation and CO Production. To evaluate the influence of the concentration of sulfur on CO production and coke formation in the reactor coil and in the TLE coking tests with addition of dimethyl disulfide (DMDS) during cracking of naphtha III were performed. All tests were performed under identical cracking conditions, as specified in Table 4 (test 3). Dimethyl disulfide was used as an additive because it is frequently applied in industrial practice. The following techniques of sulfur addition were investigated: continuous addition, presulfidation, and presulfidation followed by continuous addition. The range of amounts of sulfur used in this study is given in Table 7. In addition to influencing CO production and coke formation, sulfur compounds are also believed to influence the rate of cracking reactions by interfering with hydrogen-transfer and termination reactions.18,19 No influence on the rate of cracking and the associated product selectivities during tests with addition of DMDS

4358

Ind. Eng. Chem. Res., Vol. 40, No. 20, 2001

Table 7. Range of Amounts of Sulfur Used presulfidation continuous + continuous presulfidation addition addition S feed flow rate mg S s-1 S concentration in feed (ppmw of S) a

0.12-0.95

-

0.48-1.89

-

50-800

200-800

Reactor surface area ) 0.258 m2, TLE1 surface area ) 0.143

m2.

Figure 3. CO content in the effluent as a function of run length during cracking of naphtha III for a blank run, continuous addition of DMDS, after presulfidation with DMDS, and for presulfidation followed by continuous dosing of DMDS.

Figure 4. Influence of the amount of sulfur used during continuous addition on the amounts of coke deposited in the reactor and in the TLE1 during cracking of naphtha III. For cracking conditions, see Table 4, test 3.

was observed. The effluent composition during cracking of naphtha III under standard cracking conditions is given in Table 6 (test 3). The addition of sulfur had a marked influence on coke deposition and CO production, however. 3.2.1. Influence of Continuous Addition of DMDS. A series of tests with continuous addition of increasing amounts (50, 200, 400, and 800 ppmw of S relative to the naphtha feed) of DMDS to the feed was performed. The cracking conditions are specified in Table 4 (test 3). The addition of 50 ppmw of S as DMDS results in a 90% reduction of CO in the effluent. No further decrease of CO production with increasing quantities of DMDS was observed. Here again, a low, stable CO production is observed, and the high values at the beginning of the run are completely suppressed. Figure 3, comparing the CO production as a function of run length for a blank run with that for a run with the addition of 200 ppmw of S as DMDS, illustrates this. Figure 4 presents the amounts of coke deposited in the reactor coil and in TLE1 as function of the amount

of S added. Unlike the observed reduction in CO production, the weight of coke deposited in the reactor coil increases by 85% upon addition of 50 ppmw of S. At higher sulfur concentrations, the weight of coke deposited decreases slowly but remains higher than the value observed for a blank run. From 400 ppmw of S onward, a constant value that practically coincides with the value observed for a blank run is reached. The influence of DMDS on coke deposition in TLE1 is far less pronounced but the same trend as in the reactor coil can be seen. Upon addition of 50 ppmw of S, the rate of coke deposition in TLE1 increases by 20%. From 200 ppmw of S onward, a constant value that almost coincides with the value observed for a blank run is observed. Although data on coke formation obtained in a tubular reactor are less suited to obtaining mechanistic insights, some conclusions on the influence of sulfur on the mechanism of coke formation can be made. Kinetic studies indicate that the coking rate in the cracking coil is highest at the beginning of a run and then gradually decreases with time to reach a constant value. It is generally accepted that coke deposition in the reactor coil is a complex phenomenon and occurs through three different mechanisms: (1) heterogeneous catalytic coking, (2) heterogeneous noncatalytic coking, and (3) homogeneous noncatalytic coking. The high initial coking rate is associated with a catalytic mechanism consisting of (a) chemisorption of hydrocarbons from the gas phase on the metal wall, followed by surface reactions leading to the formation of surface carbon atoms; (b) dissolution and diffusion of surface carbon atoms through the metal particles; and (c) detachment of the metal particle from the surface with precipitation of carbon. Once the metal particles become covered with coke, their catalytic activity decreases, and growth of the coke layer occurs through reactions of coke radicals with gas-phase coke precursors. The coke radicals are generated by hydrogen abstraction from the partially dehydrogenated surface of the carbon layer formed during the catalytic stage by means of gas-phase radicals. This mechanism implies that the number of radicals in the coke matrix depends not only on the gasphase composition but also on the characteristics of the coke layer formed during the catalytic stage. Important coke precursors are radicals, unsaturated compounds (ethylene, acetylene, butadiene, etc.), and aromatics.20 This mechanism is the most important source of coke in the reactor coil, as it operates practically over the complete run length. The homogeneous noncatalytic coke formation described by Lahaye21 is of importance at temperatures above 900 °C and with heavier feeds only. As the first step in coil coking involves chemisorption of hydrocarbons, it is clear that the properties of the metal surface play an important role in the initial stage of coke deposition. Sulfur chemisorption on metal surfaces is known to modify the adsorption characteristics and the catalytic properties of metals.22,23 Depending on the amount of sulfur adsorbed, the catalytic activity of the metal can either decrease or increase.24 Therefore, factors such as the reactor wall temperature, the nature and surface state of the metal prior to the adsorption of sulfur, the nature and concentration of the sulfur compounds interacting with the metal wall, and coadsorption of steam and/or hydrocarbons all contribute to a large extent to the ultimate surface state and

Ind. Eng. Chem. Res., Vol. 40, No. 20, 2001 4359

thus to the catalytic activity of the metal wall. Moreover, the solubility and the diffusion of carbon in the metal are influenced by sulfur addition. Investigations of sulfur uptake25 by iron and steel in hydrogen sulfidecontaining atmosphere at 1000-1100 °C showed that with increasing carbon content, the rate of diffusion of sulfur decreases, while with increasing sulfur content, the rate of carbon diffusion decreases. Therefore, continuous addition of sulfur can be expected to have a complex influence on all steps involved in catalytic coke formation. Kim et al.24 observed a complex pattern for the catalytic activity of cobalt for coke formation from ethylene at 535 °C as a function of the amount of sulfur used during continuous addition. At the beginning of a run, steam, hydrogen, COx, hydrocarbons, and the sulfur additive are in contact with the preoxidized surface of the reactor wall. During preoxidation, a surface oxide layer is formed. This oxide layer mainly consists of Fe2O3, nickel oxide, chromium oxide, and mixed chromium, iron, nickel oxides. Because the alloy contains small amounts of Si and Mn, manganese oxide and silicon oxide will also be present at the surface. In naphtha cracking, the ratio of steam to hydrogen is such that chromium oxide, silicon oxide, manganese oxide, and the mixed chromium, iron, nickel oxides (spinel) on the surface of the alloy remain stable, iron is present as Fe3O4, and nickel oxide is reduced. Also, under the conditions prevailing in the cracking coil, thermal decomposition of DMDS will occur, the main decomposition products being dimethyl sulfide, methanethiol, carbon disulfide, and hydrogen sulfide.26-28 In a tubular reactor, the dependency of the concentration on p, T, and the composition of the gas phase leads to a concentration profile of DMDS and its decomposition products as a function of the axial distance in the reactor. The concentration profile of sulfur during a test with continuous addition of 800 ppmw of S was calculated taking into account the effects of p, T, and expansion. The sulfur concentration decreased from 20.9 × 10-3 mol of S/m3 at the entrance of the reactor to 8.4 × 10-3 mol of S/m3 at the reactor outlet. Clearly, the gas-phase concentration of S decreases as a function of the axial distance in the reactor. As the interaction of sulfur compounds with the surface depends on the local conditions, the amount of sulfur adsorbed on the surface will vary as a function of the axial distance in the reactor. Because sulfur adsorption can also influence dehydrogenation reactions,29 it can be expected that the characteristics of the coke layer, and in particular its hydrogen content and its microstructure (density, porosity), can be altered upon addition of sulfur. Thereby, the kinetics of the hydrogen-abstraction reactions responsible for the creation of radical centers in the coke layer can be influenced. Also, SH radicals, originating from interactions between gas-phase radicals and hydrogen sulfide, can interfere in the hydrogen abstractions and in the termination reactions of the radical coking mechanism, thereby creating additional radical centers in the coke matrix and thus enhancing its activity for further growth.5 Coke in the TLE is deposited at lower temperatures (370-700 °C); it contains more hydrogen and has a different thermal conductivity than the coke formed in the reactor coil.30 Although the mechanisms that contribute to coke formation in the TLE are still under debate, it is clear that TLE coking cannot be regarded

as a simple continuation of coil coking at lower temperatures. Coke formation in the TLE has often been attributed to physical condensation of the high-boiling cracked products on the colder metal wall. However, as argued by Ranzi et al.,30 in a properly designed TLE, the wall temperature remains higher than the dew point. Coke deposition in the TLE follows a pattern similar to that observed for coil coking: a high initial coking rate is followed by a gradual decrease to an almost constant value,31 a behavior typical of a catalytic reaction with deactivation of the catalyst. Kopinke et al.32 showed that the material on which coke deposition occurs influences the coking, thereby providing further evidence of a catalytic route to TLE coking. At the low temperatures prevailing in the TLE, the radical coke growth mechanism will not contribute substantially to TLE coking. For the effluent obtained during cracking of naphtha III under standard cracking conditions, the highestboiling component observed was dimethylchrysene. The dew point of the effluent was calculated to be 186.9 °C (P ) 170 000 Pa). As the TLE1 outlet temperature is 370 °C and the wall temperature is always higher than 300 °C, physical condensation is not expected to contribute to coke formation in TLE1 under the conditions used in this study. Therefore, coke formation in TLE1 mainly proceeds through a catalytic route. As for coil coking, the first step in TLE coking involves chemisorption of hydrocarbons on the wall of the TLE. In this case too, the amount of sulfur adsorbed on the surface will play a crucial role in determining the ultimate surface state and thus the surface reactivity toward coke deposition. Therefore, the same factors as for the reactor play an important role. At the beginning of the run, steam, hydrogen, COx, hydrocarbons, DMDS, and/or its decomposition products are in contact with the preoxidized surface of TLE1. At the temperatures prevailing in TLE1, the ratio of hydrogen to steam in the effluent is such that the components present in the surface oxide layer are the same as in the reactor. However, because DMDS is introduced at the inlet of the reactor coil and decomposes as a function of the axial distance in the reactor, the nature and concentration of sulfur compounds entering TLE1 will be different from those of the sulfur compounds entering the reactor coil. As the interaction of sulfur compounds with the surface depends on the local conditions, the amount of sulfur adsorbed on the surface varies as a function of the axial distance in the TLE. The influence of the local conditions, i.e., p and T, on the sulfur concentrations in TLE1 during the experiment with continuous addition of 800 ppm of S as DMDS was calculated. The sulfur concentration increased from 9.8 × 10-3 mol of S/m3 at the entrance of TLE1 to 11.8 × 10-3 mol of S/m3 at the outlet of TLE1. Clearly, the gas-phase concentration of S increases as a function of the axial distance in the TLE. Because the TLE temperatures are lower than the temperatures prevailing in the cracking coil, for the same concentration of sulfur compounds in the gas phase, the amount of sulfur adsorbed on the TLE surface can be expected to be higher than the amount adsorbed in the reactor. 3.2.2. Influence of Presulfidation with DMDS. To evaluate the influence of presulfidation on CO production and coke deposition, a series of tests with increasing quantities (0.12, 0.48, and 0.95 mg of S/s) of DMDS during presulfidation was performed. The detailed pre-

4360

Ind. Eng. Chem. Res., Vol. 40, No. 20, 2001

Figure 5. Influence of the amount of sulfur used during presulfidation on the amounts of coke deposited in the reactor and in TLE1 during cracking of naphtha III. For presulfidation conditions, see Table 3, tests 4-6. For cracking conditions, see Table 4, test 3.

sulfidation conditions are specified in Table 3 (tests 4-6). The cracking conditions are given in Table 4 (test 3). Figure 3 presents the CO production as a function of run length during cracking after presulfidation for 1.8 ks with 0.48 mg of S/s. The CO production steadily increases as a function of run length to reach a constant value. As compared to a blank run, the maximum at the beginning of the run is largely suppressed. However, presulfidation of the surface with DMDS is less effective in reducing CO production than continuous addition of DMDS. A 1.8 ks presulfidation with 0.12 mg of S/s decreases the CO production by 27% as compared to the value observed after a steam pretreatment. Presulfidation with higher doses of DMDS gave no further reduction of the amount of CO produced. The amounts of coke deposited in the reactor and in TLE1 as functions of the amount of sulfur used during presulfidation are presented in Figure 5. Presulfidation of the surface decreases the amount of coke deposited in the reactor by ∼35%. No influence of the amount of DMDS used during presulfidation was observed. Although somewhat less pronounced, the same trend is observed for TLE coking. Upon presulfidation, the amount of coke deposited in TLE1 decreases by ∼30%. During presulfidation, steam and sulfur compoundss DMDS and/or its decomposition productssinteract with an oxidized metal wall, and this interaction leads to a sulfided surface whose catalytic properties will crucially depend on the amount of sulfur adsorbed. The amount of sulfur adsorbed on the surface depends on the local conditions i.e., temperature, pressure, nature and concentration of sulfur compounds, and will vary as a function of the axial distance in the coil and in the TLE. The influence of the local conditions, i.e., p and T, on the sulfur concentrations in the reactor and in TLE1 during the presulfidation with 0.95 mg of S/s as DMDS was calculated. In the reactor, the sulfur concentration increased from 8.2 × 10-3 mol of S/m3 at the reactor entrance to 5.5 × 10-3 mol of S/m3 at the reactor outlet. In TLE1, the sulfur concentration increased from 6 × 10-3 mol of S/m3 at the inlet to 7.2 × 10-3 mol of S/m3 at the outlet TLE1. Clearly, in the reactor, the gas-phase concentration of S decreases as a function of the axial distance in the reactor during presulfidation, whereas it increases as a function of the axial distance in the TLE. At the beginning of a run, steam, hydrogen, COx, and hydrocarbons are in contact with a presulfided surface. Because no influence of the amount of sulfur used

Figure 6. Amount of coke deposited in the reactor and in TLE1 during cracking of naphtha III for a combination of presulfidation with DMDS followed by continuous dosing of DMDS. For presulfidation conditions, see Table 3, tests 7-9. For cracking conditions, see Table 4, test 3.

during presulfidation on the rate of coke deposition was observed, it could be concluded that, from a dosing of 0.12 mg of S/s onward, the metal surface is completely saturated with sulfur. However, the behavior of CO production as a function of run length seems to suggest that sulfur is gradually removed from the surface. 3.2.3. Influence of Presulfidation Combined with Continuous Addition of DMDS. To evaluate the influence of presulfidation followed by continuous dosing of DMDS on CO production and coke deposition, a series of tests with addition of increasing quantities of DMDS (0.48, 0.95, and 1.89 mg of S/s as DMDS) during presulfidation followed by continuous dosing (200, 400, and 800 ppmw of S as DMDS relative to the hydrocarbon feed) was performed. The detailed presulfidation conditions are specified in Table 3 (tests 7-9). The cracking conditions are given in Table 4 (test 3). A 1.8 ks presulfidation with 0.48 mg of S/s followed by the continuous addition of 200 ppmw of S results in a 90% reduction of CO in the effluent as compared with a blank run. No further decrease of CO production with increasing quantities of DMDS was observed. Figure 3 presents the CO production as a function of run length after 1.8 ks presulfidation with 0.48 mg of S/s followed by continuous addition of 200 ppmw of S as DMDS. Our observations indicate that, even after presulfidation, continuous dosing of DMDS during cracking is required to obtain a low and stable CO production as a function of run length. Figure 6 indicates that the addition of increasing amounts of DMDS both during presulfidation and during continuous addition reduces coke deposition in the coil for up to 400 ppmw of S. Further increasing the amount of DMDS restores coke formation to the level observed for a blank run. A 1.8 ks presulfidation with 0.95 mg of S/s followed by a continuous dosing of 400 ppmw of S decreases the coke deposition in the coil with 15% as compared to a blank. Upon increasing the amount of sulfur during presulfidation to 1.89 mg of S/s and the continuous dosing to 800 ppmw of S, the amount of coke deposited in the coil practically coincides with that observed for a blank run. As compared to a blank run, no marked influence on the rate of coke deposition in TLE1 was observed. In summary, the following statements can be made: (1) Sulfur controls CO production very well but the amount of CO produced bears no direct relation to the

Ind. Eng. Chem. Res., Vol. 40, No. 20, 2001 4361

amount of coke formed. Thus, monitoring of the CO concentration in the effluent is not a reliable indicator for coke deposition. Although presulfidation reduces CO production as compared to an oxidizing pretreatment, continuous dosing of DMDS is necessary to obtain a low and stable CO production. (2) On a preoxidized coil surface (Incoloy 800HT), continuous dosing of DMDS in concentrations lower than 400 ppmw of S increases the amount of coke deposited. The negative influence decreases with increasing quantities of DMDS added. From 400 ppmw of S onward, no marked influence as compared with a blank run is observed. (3) Presulfidation of the reactor coil (Incoloy 800HT) with 0.12-0.95 mg of S/s as DMDS reduces the rate of coke deposition in the reactor by 35% as compared with a preoxidizing pretreatment. In this range, no dependence of coke formation on the amount of sulfur is observed. (4) Combined presulfidation and continuous addition of DMDS leads to a decrease in coke deposition in the reactor coil (Incoloy 800HT) up to 400 ppmw of S. Further increasing the amount of DMDS restores coke formation to the level of a blank run. (5) On a preoxidized TLE surface (Incoloy 800HT), continuous addition of DMDS in doses lower than 200 ppmw of S increases coke formation. From 200 ppmw of S onward, no marked influence as compared with a blank run is observed. (6) Presulfidation of the TLE surface (Incoloy 800HT) with 0.12-0.95 mg of S/s as DMDS reduces the coke formation by 30% as compared with a preoxidizing pretreatment. In this range, no dependence of the coke deposition on the amount of sulfur is observed. (7) Combined presulfidation and continuous addition of DMDS has no marked influence on coke deposition in the TLE (Incoloy 800HT). The results of this pilot-plant study clearly indicate that the influence of sulfur addition on coke formation in the reactor can strongly differ from its influence on coke formation in the TLE. In the radiant coil as well as in the TLE, the observed influence of sulfur addition is complex and strongly depends on the technique used. The amount of sulfur additive has a far more pronounced influence when it is added continuously with the feed than when it is used as a presulfidation agent. Because the data obtained in the pilot unit are average coking rates, they are less suited to obtaining mechanistic insights. However, the present observations clearly suggest that sulfur has a complex influence on all steps involved in coke formation. Coke deposition in the presence of sulfur crucially depends on the amount of sulfur adsorbed as the chemical nature and structure of the surface, and thus its activity for coke formation, are largely determined by the amount of sulfur adsorbed on the surface. Moreover, the adsorption of sulfur strongly depends not only on the concentration and the chemical nature of the sulfur compound but also on the wall temperature and the nature and concentration of the other compounds, i.e., hydrogen, COx, steam, and hydrocarbons, present in the gas phase. Further, under the conditions of thermal cracking, DMDS decomposes, and in a tubular reactor, this gives rise to a concentration profile of DMDS and its decomposition products. Therefore, to obtain more insight into the influence of sulfur on the mechanism of coke formation, studies should be performed in a microreactor with complete

mixing of the gas phase, i.e., operating under point conditions of partial pressures and temperature, equipped with an electro-balance to monitor the kinetics of coke deposition in the presence of sulfur. Extrapolation of the results obtained under point conditions to the industrially used tubular reactors then requires knowledge of the kinetics of the thermal decomposition of DMDS and the influence of the process conditions on the amount of sulfur adsorbed on the surface. Conclusions A pilot unit with a cooling section that allows for the study of coke formation in the reactor as well as in the TLE was successfully operated. Sulfur addition during thermal cracking of naphtha controls CO production very well. However, there is no direct relation between the amount of CO produced and the amount of coke formed. Thus, monitoring of the CO concentration in the effluent is not a reliable indicator for coke deposition. Although presulfidation reduces CO production as compared to an oxidizing pretreatment, continuous addition of DMDS is necessary to obtain a low and stable CO production. The influence of sulfur addition on coke formation in the reactor coil can strongly differ from its influence on coke formation in the TLE. In the reactor coil as well as in the TLE, the observed influence of sulfur addition is complex and strongly depends on the technique used. Acknowledgment Financial support from ELF AQUITAINE is gratefully acknowledged. I. Dhuyvetter gratefully acknowledges the Nationaal Fonds voor Wetenschappelijk Onderzoek for the Grant N.F.W.O.-Petrofina. Literature Cited (1) Santiago, J. A.; Francesconi, J. D.; Moretti, N. L. Controlling CO and CO2 cuts downstream processing costs. Oil Gas J. 1983, 81 (39), 78-82. (2) Goossens, A. G.; Dente, M.; Ranzi, E. Improve steam cracker operation. Hydrocarbon Process. 1978, 57, 227-236. (3) Taylor, D. M.; Allen, L. M. Amoco Anticoking Technology, A process for inhibiting coking in pyrolysis furnaces. In Proceedings of the 6th Ethylene Producers’ Conference; American Institute of Chemical Engineers: New York, 1994. (4) Woerde, H. M.; van den Oosterkamp, P. F.; Brun, C.; Sposisto, L. New developments in fouling inhibition of cracking coils and TLE’s. Presented at the Technip Group Symposium, Noordwijk, The Netherlands, Oct 1999. (5) Reyniers, M.-F.; Froment, G. F. Influence of metal surface and sulfur addition on coke deposition in the thermal cracking of hydrocarbons. Ind. Eng. Chem. Res. 1995, 34, 773-785. (6) Reed, L. E. The effect of sulfur compounds and Phillips antifoulants in ethane pyrolysis. Presented at the Symposium on Coke Formation and Mitigation, Division of Petroleum Chemistry, 210th National Meeting, American Chemical Society, Chicago, IL, Aug 20-25, 1995. (7) Bajus, M.; Vesely, V. Pyrolysis of hydrocarbons in the presence of elemental sulfur. Collect. Czech. Chem. Commun. 1980, 45, 238. (8) Bajus, M.; Baxa, J. Coke formation during the pyrolysis of hydrocarbons in the presence of sulfur compounds. Collect. Czech. Chem. Commun. 1985 50, 2903. (9) Depeyre, D.; Flocoteaux, C.; Blouri, B.; Ossebi, J. G. Pure n-nonane steam cracking and the influence of sulfur compounds. Ind. Eng. Chem. Process Des. Dev. 1985, 24, 920-924. (10) Kolts, J. H. Heterogeneous and homogeneous effects of H2S on light hydrocarbon pyrolysis. Ind. Eng. Chem. Fundam. 1986, 25, 265.

4362

Ind. Eng. Chem. Res., Vol. 40, No. 20, 2001

(11) Trimm, D. L.; Turner, C. J. The pyrolysis of propane: 2. Effect of hydrogen sulfide. J. Chem. Technol. Biotechnol. 1981, 31, 285-289. (12) Velenyi, L. J.; Yihhong, S.; Fagley, J. C. Carbon deposition in ethane pyrolysis reactors. Ind. Eng. Chem. Res. 1991, 30, 17081712. (13) Van Damme, P. S.; Froment, G. F. Putting computers to work; Thermal cracking computer control in pilot plants. Chem. Eng. Prog. 1982, 78, 77-82. (14) Dierickx, J.; Plehiers, P. M.; Froment, G. F. On-line gas chromatographic analysis of hydrocarbon effluents: Calibration factors and their correlation. J. Chromatogr. 1986, 362, 155-174. (15) Wang, X. L.; Gomez, M. F.; De Saegher, J. J.; Froment, G. F.; Woerde, H. Thermal cracking of i-butane at high pressure. Oil Gas Eur.Mag. 1995, 4, 20-21. (16) Rostrup-Nielsen, J. R. Steam Reforming Catalysts: An Investigation of Catalysts for Tubular Steam Reforming of Hydrocarbons; Teknisk Forlag: Copenhagen, Denmark, 1975. (17) Rostrup-Nielsen, J. R. Sulfur passivated nickel catalysts for carbon free steam reforming of methane. J. Catal. 1984, 85, 31-34. (18) Rebick, C. H2S catalysis of n-hexadecane pyrolysis. Ind. Eng. Chem. Res. 1981, 20, 54-59. (19) Scacchi, G.; Dzierszynski, M.; Martin, R.; Niclause, M. H2S inhibition of the pyrolysis of ethane. Int. J. Chem. Kinet. 1970, II, 115-122. (20) Kopinke, F.-D.; Zimmerman, G.; Reyniers, G.; Froment, G. F. Relative rates of coke formation from hydrocarbons in steam cracking of naphtha: 2. Paraffins, naphthenes, monoolefins, diolefins, and cycloolefins, and acetylenes. Ind. Eng. Chem. Res. 1993, 32, 56-60. (21) Lahaye, J.; Badie, P.; Ducret, J. Mechanism of carbon formation during steam cracking of hydrocarbons. Carbon 1977, 15 (2), 87-99. (22) Kiskinova, M. P. Electronegative additives and poisoning in catalysis. Surf. Sci. Rep. 1988, 8, 359-402.

(23) Goodman, D. W. Chemical modification of chemisorptive and catalytic properties of nickel. Appl. Surf. Sci. 1984, 19, 113. (24) Kim, M. S.; Rodriguez, N. M.; Baker, R. T. K. The interplay between sulfur adsorption and carbon deposition on cobalt catalysts. J. Catal. 1993, 143, 449-463. (25) Bramley, A.; Haywood, F.; Cooper, A.; Watts, J. Diffusion of nonmetallic elements in iron and steel. Trans. Faraday Soc. 1935, 31, 707-734. (26) Brun, C. Presented at the ATOFINA Seminar, Sulfur Chemicals in Oil & Gas Industries, Ramsar, Iran, Sep 26-27, 2000. (27) Kroto, H. W.; Suffolk, R. The photoelectron spectrum of an unstable species in the pyrolysis products of dimethyldisulphide. Chem. Phys. Lett. 1972, 15, 545-548. (28) Bock, H.; Mohmand, S. Instabile Zwischenprodukte in der Gasphase: Der thermische Zerfall van Alkylsulfiden RSnR. Angew. Chemie 1982, 89, 105-106. (29) Somorjai, G. A. On the mechanism of sulfur poisoning of platinum catalysts. J. Catal. 1972, 27, 453-456. (30) Ranzi, E.; Dente, M.; Pierucci, S.; Barendregt, S.; Cronin, P. Coking simulation aids on-stream time, Oil Gas J. 1985, 4952. (31) Bach, G.; Zimmermann, G.; Kopinke, F.-D.; Barendregt, S.; van den Oosterkamp, P.; Woerde, H. Transfer Line Heat Exchanger fouling during pyrolysis of hydrocarbons: 1. Deposits from dry cracked gases. Ind. Eng. Chem. Res. 1995, 34, 11321139. (32) Kopinke, F.-D.; Bach, G.; Zimmerman, G. New results about the mechanism of TLE fouling in steam crackers, J. Anal. Appl. Pyrolysis 1993, 27, 45-55.

Received for review December 30, 2000 Revised manuscript received May 25, 2001 Accepted June 1, 2001 IE001131B