- Email: [email protected]
Control of coking
137
Control of Coking D. L. TRIMM School of Chemical Engineering and Industrial Kensington, N.S. W. 2033 (Australia)
Chemistry,
University
of New South Wales, P, 0. Box 1,
(Received August 5, 1983)
Abstract The formation of coke during the large-scale processing of hydrocarbons is largely unwelcome. Not only does it represent a loss of product but it also leads to decreased heat transfer and to blockage. At high temperatures, carburlzation of metallic materials can occur, leading to catastrophic failure. The origins of different types of coke are discussed, and coke deposition is shown to be a balance between coke formation and coke removal. Carburization can occur when coke is deposited on metal surfaces. Means of minimizing coking and carburlzation are considered in terms of the operating parameters open to control and in terms of choice of materials. If coke does accumulate, the means of coke removal are presented, with particular attention being focused on the need to control carefully the conditions of decoking. 1. Introduction The industrial processing of organic feedstocks to produce desired chemicals involves a variety of reactions carried out over a wide range of conditions. The reactions may be catalytic or noncatalytic, may be in the liquid phase or the gas phase, and may be carried out at high or low temperature or pressure. Each system presents individual challenges to the chemist and chemical engineer but some factors are common to most. Of these, coking is of major importance since it often dictates the overall economics of the process. Coking is a collective noun, describing the accumulation of carbonaceous residues in the reactor or in the plant. The nature of the residue varies, from polymeric deposits favoured at low temperatures to carbons produced under more extreme conditions. The residues deposit on different surfaces throughout the plant, leading to deactivation of catalysts, mass transfer problems, decreased heat transfer and increased pressure drop. They may be removed mechanically or by gasification, but this entails interruption of normal plant operation and may involve degradation of the catalyst or the reactor material. As a result, there is a strong incentive to minimize the problem and to optimize conditions for removal. Removal of coke is a reasonably straightforward procedure, with care being necessary to avoid damage to a catalyst by, for example, overheating. Minimization of coke formation requires more detailed knowledge of the origins of coke since these can vary markedly. In general, coke formation is a polymerization/dehydrogenation process which is favoured by high temperatures and low pressures; in particular, the chemical reactions producing
coke can be very different. As a result, it is convenient to begin this article by considering the nature and origins of different types of coke. It should be remembered that coking is an undesired side reaction occurring concurrently with a desired main reaction. As such, process conditions are designed to minimize the former and to maximize the latter. However, it is the conditions appropriate to the desired reaction which are more important since these create boundary conditions within which coking should be minimized. Since coke is formed in many different processes, the operating conditions involved also vary widely. As a result, it is not surprising that both the nature and the origin of coke vary markedly. In this article, attention is focused first on the different types of coke and where they come from. The problems that they can cause are then discussed. Control of coking is considered and, if this falls, removal of the accumulated deposit and the problems arising during this are discussed. 2. Nature and origins of coke 2.1. Low temperature
liquid phase systems
Formation of a polymeric residue which has many similarities to coke can be observed at quite low temperatures in some systems. Thus, for example, liquid phase hydrogenation of oils can lead to deactivation of the catalyst by coking [I, 21, and pipeline transfers of hot oils can lead to accumulation of residues in regions of stagnant flow [2]. The rates of residue formation are low, but the deposit appears to originate from polymer-
Gem. Eng. F’roces.. 18 (1984) 137-148
0 Elsevier Sequoia/Printed
in The Netherlands
138 Polymerization can be minimized by adding a free radical capping agent such as supplementary hydrogen [3,41> .
ization of olefinic, naphthalenic and aromatic compounds, followed by slow dehydrogenation and rearrangement reactions to form stable coke deposits. The reactions involved, being slow, are open to question, but it is believed that the general type of process involved can be represented by the sequences below [3,4]. For clarity, the feedstock is considered to be butadiene.
<
+
a,- a
oj
+ 0
(5)
but the stability of some of the larger free radicals makes this process less efficient than would be expected. As a result, coke formation can be reduced but not eliminated. 2.2. Gas phase coke
+2Hz
(2)
+ H’
(3)
.
+ CzH4
+Hz-2
2
-
Q
-
+ H’
(4)
3 co Free radicals are believed to be important in the processes and, at low temperatures, resonance stabilized free radicals based on aromatic compounds
can be relatively long lived [3,4]. Their concentration is low, but their long lifetimes increase the chance of molecular weight build-up by polymerization, particularly in stagnant zones in the reactor or pipeline. These slow polymerization reactions increase molecular weight and decrease solubility, causing precipitation on any convenient surface. Once precipitated, the compounds are not necessarily inert, and further liquidsolid free radical polymerizations can occur [5] leading to growth of the deposit. The material can also rearrange to form a more ordered structure (usually involving elimination of hydrogen) but this is slow at low temperatures.
Coke formation at higher temperatures is more important and can occur in the gas phase or on a surface. The importance of either depends on the reactivity of the coke producing gas and the nature of the surface but, in general, gase phase coking is more important at higher temperatures (L 600 “C). Gas phase carbons are believed to be produced by polymerization/dehydrogenation processes, and the deposits are known to contain a range of polycyclic aromatic materials together with carbon. The carbon (gas phase carbon) is generally disordered (Table 1) and has the appearance shown in Fig. 1. The polycyclic aromatics (tars) are difficult to distinguish from carbon since they have the same appearance and have a high carbon:hydrogen ratio. They can be dissolved in suitable solvents for analysis, and this shows that components ranging from naphthalene (two aromatic rings) to hexabenzotriphenylene (ten aromatic rings) are present. Calculations of the vapour pressures of these compounds show that they can exist in the solid phase, even at temperatures up to 1000 “C, and, as a result, will behave much as carbon itself. There seems little doubt that tars and carbon are produced by free radical reactions similar to those described above. Studies of gas phase reactions show that, with increasing residence time, gas phase products, tars and gas phase carbon are produced in turn and that tars and carbons have a well-defined induction period (Fig. 2) [6]. As a result, it is logical to suggest that the feedstock produces monomers that polymerise to greater extents over longer times, eventually forming first tars and then carbon. Undoubtedly this does happen, but it is not the complete explanation, since tars and carbon can be formed at exactly the same residence time in some systems [7, 81. We shah see that
TABLE 1. Typical properties of carbons
Spe
Typical formation temperature CC)
C:H ratio
Lc (A)
LA (A)
d (A)
Gas phase Surface Catalytic Qaminar) Graphite
1000 1000 600 2000
8 >80 20
12.2 28 200
42 51 400 OD
3.6-3.7 3.45-3.55 3.4 3.35
The data refer to crystals in which Lc is the length and LA the breadth. The crystals contain layers of carbon atoms parallel to each other at distance d apart.
Fig. 1. Photographs of typical coke.
volved are mainly dehydrogenation and ordering, the latter process involving a reorganization of randomly deposited coke to give, as a final product, an ordered graphite-like structure. Complete ordering rarely occurs and the carbon may, or may not, have a distinctive appearance (Fig. 1). If the original deposit is gas phase carbon, a liiited amount of ordering does not lead to a major change in morphology (Fig. 1); if the original deposit is tar, the surface carbon will have a higher degree of long-range order. 2.3. Cattilyticcoke
Fig. 2. The production of material during steam cracking as a function of residence time.
the induction period for tar and carbon production has important ramifications for other aspects of coking. Tars and gas phase carbon will collect on any convenient surface and, even if the surface is non-catalytic, further reactions will occur to produce a more ordered deposit of surface carbon (Table 1). The reactions in-
Coking is also affected by the presence of a catalyst such as a transition or noble metal (catalytic carbon). Highly crystalline deposits of carbon can be formed at relatively low temperatures (-400 “C) and no tars are produced. In addition to metal carbides, three types of carbon are seen-laminar graphite, non-oriented carbon and whisker carbon (Fig. 3). Laminar graphite tends to form under conditions of high temperature and low hydrocarbon pressure, while non-oriented carbon is favoured under conditions that favour high supersaturation of carbon in the metal. Carbon whiskers are produced mainly on transition metals and can be up to 7 pm long and 0.1 w wide.
ably well understood, but they may be complicated by mass and heat transfer effects (in reactions at high pressure, for example.) Coking, in these systems, results mainly from acid catalysed polymerization reactions of carbonium ions [ 121. Reactions such as cracking involve the formation of carbonium ions from olefins: RHCH=CHR’H + H+A- e RHCH,-C+HR’H S H,;R’H
t A- e
RHCH,-&HR’H
t A-
R=CH2 + HzC+R’H
(6) (7)
(8)
H+A- + R’=CH?
The carbonium ions may react with other olefins or with aromatics to give higher molecular weight species and, eventually, carbonaceous deposits: RCH,-:HR’
t RCH=CHR’ e
RCH2-FHR’
(9)
R&H-CHR’ t Fig. 3. Catalytic carbons: A, whisker carbon; B, non-oriented
carbon; C, graphite carbon.
RCH,--dHR’
The production of catalytic carbon can best be described with the aid of Fig. 4 [9]. Hydrocarbons adsorb on a metal surface and may react to produce gas phase products or dehydrogenated adsorbed intermediates. This process continues until carbon is produced on the surface which, in turn, can isomerize to other forms of carbon [5,6]. Depending on the metal, up to four forms of carbon have been identified ranging from o-carbon (very reactive), &carbon (amorphous and less reactive) and crystalline carbon to metal carbides of varying stoichiometry [IO]. The subsequent fate of this carbon determines the nature and amount of coke formed. o-carbon may be gasified or may react to produce P-carbon. &carbon or crystalline carbon may be gasified, but are more likely to encapsulate the surface, leading to laminar graphite or non-oriented carbon. Alternatively they may dissolve in the metal, diffuse to a grain boundary and precipitate out, producing metal carbides and whiskers of carbon with metal particles at their tips [9, 111. This process weakens the metal (see later). The relative amounts of different types of carbon then depend on the production of the intermediate (which can be reduced by efficient gasification), on the balance between the different forms of carbon and on the balance between carbon dissolution and encapsulation. Coke may also be formed on non-metallic catalysts and coking of acidic oxides and sulphides is well established. The underlying reaction mechanisms are reasonHydrocarbon
tl Dissolved carbon
Fig. 4. Model for catalytic carbon production.
Dissolved carbon
+0
+ A- ~~~H-H~~H+A_
(10) Dehydrogenation and isomerization will then occur but the resulting coke is less well ordered in its final form as a result of the mechanism of formation. Coke formation on sulphides appears to result from several causes and is of particular interest because these catalysts are often used to hydrotreat heavy oils [l, 21. The high molecular weight of the feedstock favours coke formation and excess hydrogen is used in an attempt to reduce the problem [ 11. The mechanism of coking may involve ionic and non-ionic intermediates, and it is difficult to tell which process may be more significant. Mass transfer processes may also influence coking, since hydroprocessing of heavy oils is often a gasliquid-solid reaction. Transfer of hydrogen through the liquid may be too slow to stop coking and intermediates may be trapped in pores to produce a local concentration excess that favours coking. It is seen that the formation of coke is a complex process that can involve many different processes. Before considering how these may be minimized, it is useful to recognize what kinds of problem can arise as a result of coke formation. 3. Problems caused by coking Most industrial chemical processes involve a catalyst, and it is not surprising that catalyst deactivation as a result of coke formation is perceived to be a major problem. The effect of different forms of coke on the kinetics of complex reactions has received much attention [ 131 and the deactivation/regeneration cycle is of major importance in determining the overall economics of a process [14]. There is little doubt that minimizing catalyst coking is a matter of considerable interest.
141 Coking is of major interest to chemical engineers, however, for other reasons, and these may be illustrated by reference to a flow diagram of a steam cracking plant (Figs. 5 and 6) [ 151. The process involves pyrolysis of a light feedstock (usually naphtha, ethane or propane) in the presence of steam to produce light olefins and other valuable products. The feedstock is mixed with steam and passed to a furnace countercurrent to the flue gas flow (Fig. 5). The gases pass through a convection heating section and a radiant coil before leaving the furnace to a transfer line heat exchanger, a water quench and a separating train (Fig. 6). The outputs from two furnaces are often combined in a bobbin before passage to the transfer line exchanger (TLE). Typical operating conditions involve a residence time in the radiant zone of 0.2-0.55 s at a pressure of 0.150.2 X lo6 Pa. The steam-hydrocarbon ratio in the feed depends on the nature of the hydrocarbons, varying from 0.25-0.4 (weight basis) for ethane to 0.8-1.0 for gas oil. The temperatures in the convection and radiant zones are about 600 and 900 ‘C respectively, and the specific heat requirements are high (-5000 kJ/kg for
naphtha-steam). As a result, high heat transfer rates are required. The gases that leave the furnace contain significant quantities of heat and much of this can be recovered by use of a TLE in which the gas temperature is dropped from around 900 “C to about 450 ‘C. The gases themselves contain a mixture of valuable components which must be separated, dried and purified. A typical processing train is shown in Fig. 6. After cooling in the TLE, residual free radicals are deactivated and some unwanted products removed in the oil quench. Heavy ends are then separated by fractionation and acid gases removed from the light ends. After drying, subsequent separation gives C1, CZ, C3 and C4 streams together with residual naphtha that can be used as gasoline. Acetylenic products are a problem in the C&C,, streams and must be removed, usually by selective hydrogenation. Separation of the remaining C2 and C3 products yields the major products, ethylene and propylene, the corresponding paraffins being recycled to the cracker. This type of plant layout is typical of many processes, with the exception that the temperature is high (-900 ‘C in the radiant zone) and that there is no catalyst. It is a convenient reference, however, to illustrate other problems caused by coke formation. The first of these involves materials of construction. The length and diameter of the reactor in the radiant zone are 90-120 m and 9-10 cm respectively. Because of the high heat transfer requirement, the tubes are made from heat resistant nickel-chromium alloys and coking catalysed by these alloys has been found to occur. The temperatures in the radiant zone are about 900 ‘C, and gas phase coking produces tars, gas phase carbon and, after deposition and rearrangement, surface carbon. Tars may deposit in the reactor or in the
TRINSFER UNE HEAT EXL”Plmjl
Fig. 5, Reactor/furnace arrangements for the steam craker.
ETH4NE OR PROPANE OR NAPHTHA OR ^.^ ^. STEAM CRACKER FUEL OIL I
ETHANEIPROPANE
I I
“2 RESIDUE
CZH2
REMOYAL
ETHYLENE FRACTION
GAS
ETHYLENE
I
PROPYLENE I
) Fig. 6. Typical processing train for steam cracking.
PYROLYSIS GASOLINE
142
bobbin/TLE, where a sharp drop in temperature occurs. Because of the induction period for the production of tars and carbon (see above) the convection zone and the early part of the radiant zone of the tube remains relatively coke free, but the back end of the tube, the bobbin and the TLE coke heavily. A typical deposit on a tube is shown in Fig. 7, and it is obvious that the thick layer of coke will increase significantly the heat transfer across the tube. As a result, the second problem arising from coking is due to decreased thermal efficiency. Coke deposited on a tube is not inactive, but diffuses, in time, into the tube material. This occurs preferentially at grain boundaries in the metal, leading to carburization (Fig. 8) and to higher thZr normal stress in the noncarburized parts of the tube. Creep often becomes critical, particularly as the metal temperature may be close to 1100 ‘C (to maintain the gas temperature at around 900 “C). Carburization of the alloys, then, is the third problem caused by coke. Once a thick deposit of coke has accumulated, it is necessary to decoke the tube by gasification with a steam-air mixture. This involves taking the tube out of service and cycling the temperature of the tube to achieve even burn-off. The expansion/contraction of the tube, together with gasification of carbon in the grain
Fig. 8. Carbon penetration tubes.
Fig. 7. Coke deposits on industrial steam cracker reactor tubes (a) and transfer line heat exchanger tubes (b). (Photographs courtesy of ICI Australia Pty. Ltd.)
,
of alloys used in steam cracking
boundaries, can lead to extensive metal dusting [16] (metal drops from the surface of the tube) and to production of a disrupted surface of composition differrent from that present originally. Subsequent re-coking will be accelerated by the surface roughness [ 17,181 and, possibly, by removal of the original layer of protective material on the surface (see below). This problem can be so acute that the life of a tube may be affected more by the number of decoking-recoking cycles than by extended operation under steady state conditions. Problems in the bobbin/TLE area are different but are also very significant. Although carburization can occur, tar condensation and reorganization is more important since this may lead to very bulky deposits and to physical blocking of the TLE tubes (Fig. 7(b)). The net effect of these problems is seen not only in decreased thermal efficiency or in metal dusting but also in catastrophic failure of tubes. Photographs of some failures observed during plant operation are shown in Fig. 9. In general, it is seen that coke formation can occur in the liquid or the gas phase, in the absence or presence of a catalyst, and at high and (relatively) low temperatures. It causes problems as a result of catalyst deactivation, of increased heat transfer resistance, of carburization/corrosion of reactor tubes and of physical blocking of equipment. As a result, there is considerable interest in minimizing coke formation: how this may be done is the subject of the text below.
143 from a series of free radical reactions that can be represented as [3,4, 191 R-R’
xR’ -
-
R’ t R”
(11)
coke
(12)
High temperatures and high pressures favour these types of reaction and the first essential is to keep temperature and pressure as low as possible. If this cannot be done, residence time in the gas phase at high temperatures should be minimized. This requires careful design not only of the reactor but also of preheaters, riser space and heat exchangers after the bed. The latter is particularly important in view of the fact that a known induction period exists for the production of tars and coke (Fig. 2). A second route to minimizing coking involves dilution. Coking is a polymerization reaction and, if the free radicals can be kept apart, coking will be minimized. As a result, there are advantages in using an inert diluent (nitrogen) or a reactive diluent (hydrogen or steam). The action of these last two gases also involves free radical capping. If reactions such as
(b)
Fig. 9. Industrial steam cracker tube failures. courtesy of ICI Austrialia Pty. Ltd.)
4. Control
(Photographs
R’+H,-RHtH’
(13)
R’-+HzO-RH+OH’
(14)
occur, the coke forming intermediates are replaced by low molecular weight free radicals, whose interaction will not form coke. In addition, gasification of coke or coke precursors may occur; this may be represented as C+2Hz-CH4
(15)
C+HzO-COtH,
(16)
of coking
Coking results from an accumulation of carbonaceous deposits from the gas phase or on a catalytic surface. In the present context, it is useful to consider minimization of gas phase coke formation, minimization of catalytic coke formation and minimization of deleterious effects due to coking. Since it is usually impossible to avoid coking altogether, a final section deals with coke removal. It is important to remember that the accumulation of coke results from a balance between coke deposition and coke (or coke precursors) removal. Control of coking then depends on minimizing deposition and maximizing gasification-either of coke or its precursors. This is distinct from coke removal during regeneration, which is a separate process involving interruption of normal operations. It is also necessary to remember that coking is an unwanted side reaction that occurs concurrently with a desired main .reaction. Any measures designed to minimize coking must take into account the operating conditions required for the desired reaction. ‘4.1. Control of coke from the gas phase Gas phase coking is dependent primarily on the nature of the gas and the operating conditions. It results
Reactions (13) and (14) can be considered as free radical deactivating reactions, and similar effects may be achieved in other ways. Thus, for example, free radical chains may be terminated at a surface [20] H’+H’+M-H2
tM
(17)
and, if the free radical hits a surface before it builds ulr size, the coking sequence can be interrupted before coke is formed.-As a result, there is an incentive to increase the surface:volume ratio in order to minimize gas phase free radical reactions. Of course, it is necessary that the pressure drop should not be too large and that the surface introduced should not be catalytic, but this may be achieved with relative ease. Similar effects can be achieved with gas phase additives, but these are rarely used in practice. Free radical termination reactions require removal of energy, either at a surface or to a molecule that can adsorb the energy. Fundamental studies use, for example, sulphur hexafluoride [20,21], but these materials are impracticable in an industrial situation. There is evidence that sulphur containing molecules present in a typical feed may act in the same way, but this is far from proven.
144 Control of gas phase coking then involves careful control of temperature and pressure, increasing dilution with an inert or a gasifying/free radical capping gas and optimizing surface:volume ratios in the heated space. 4.2. Control of catalytic coking The mechanism of coke formation on catalytic surfaces depends on the nature of the feed, of the catalyst and on operating conditions. Although the mechanism may differ, it turns out that means of controlling coke are very similar. In the first instance, these may be discussed with reference to the coking of metals. The process of coking is summarized in Fig. 4. Hydrocarbons adsorb on a surface to produce one or more intermediates. These may react to form the desired products or to form coke. The latter reaction, in the case of platinum, for example, involves the migration of an intermediate across the surface to a nucleation site where polymerization occurs to form a partially hydrogenated deposit [S]. This then reacts further, by dehydrogenation/isomerization, to form coke. On some metals (nickel, iron, cobalt, etc.) the coke may then remain on the surface or may dissolve in the metal 19,111. On the basis of this sequence, coke could be controlled by interference in several steps: (a) Temperature and pressure could be controlled so that the thermodynamic driving force for coking is minimal. (b) The intermediates on the surface could be controlled so that those involved in the desired reaction are maximized and those involved in coking are minimized. (c) Coke forming intermediates could be deactivated, for example by gasification. (d) Migration of coke precursors across the surface could be restricted. (e) Polymerization/nucleation could be minimized. (f) The partially hydrogenated deposits could be gasified rather than allowed to dehydrogenate. (g) Dissolution in the metal could be minimized. This is found to be more relevant to the following section, and is considered there. Control of temperature and pressure depends on the thermodynamics of the desired reaction, and is particular to a system. For any process, temperature-in particular-should be minimal although, as a catalyst deactivates, temperature is often increased to maintain production. This will, in turn, increase coking and lead, eventually, to complete deactivation. It may or may not be possible to control intermediates on the surface since a common intermediate may be involved in both the desired reaction and in coking. Methanation on nickel is a case in point [lo]. Carbon monoxide adsorbs to produce o-carbon, P-carbon, crystalline carbon and nickel carbide. o-carbon may react with hydrogen to form methane or may isomerize to form P-carbon. p-carbon is less reactive and rearrangement to crystalline carbon and coke is favoured. As a result it is impossible to control the nature of the intermediates-what must be done is to control the further reactions of o-carbon.
In other systems, it may be possible to control the intermediate. Ethylene, for example, may adsorb as a pi-bonded species or as a sigma-bonded species [22], with hydrogenation being favoured by the former and coke formation by the latter. In this case, it is obvious that we should choose a catalyst or adjust the catalyst to favour pi-bonded intermediates. Exactly how this is done is not easy to explain, since catalyst designers do not always understand what is going on with a particular catalyst. In some cases, however, it is possible to design a catalyst in this way. Coking involves migration over the surface and polymerization, and it has been found to be possible to control these processes by controlling the geometry of the surface. The principle of the method involves limiting the number of near-neighbour catalytically active sites (ensemble size [23]). Using methanation as an example, the desired reaction may be written in a simplified form as co-c+0
xx xx
(18)
Hz-2; 0+2Hx-H,0+4X xx &‘45f
-CH,+6X
while coking results from reactions such as 2xcx-
c-c xx xx
(22)
It is obvious that the ensemble size for coking must be much larger than that required for the main reaction and, as a result, it could be possible to limit the ensemble size to the point where the desired reaction proceeds and coking does not. At least in the case of nickel, this has been found to be possible by the addition of small amounts of a sulphur containing gas to the feed [24] _Too much sulphur poisons the catalyst, but a trace of sulphur adsorbs on nickel in a regular array which depends on the crystal face involved and the concentration of sulphur [25]. It turns out that the number of sites left uncovered is sufficient to allow the desired reaction to proceed, while not allowing coke formation. As a result, pilot plants have been run with small traces of sulphur in the feed to produce excellent results over long periods in the absence of coking [24]. The role of adsorbed sulphur may also be to reduce surface migration. Coke precursors migrate across a surface either by a desorption/re-adsorption mechanism or by movement in the adsorbed state [26]. In the latter case, the adsorbed species is loosely bonded to the surface and, if the nature of the surface changes (for example, at the boundaries of the ensemble where nickel is replaced by nickel sulphlde), then migration may be limited. The possibility of controlling coke by controlling ensemble size has been recognised only very recently [23,24], and the ramifications are still being explored.
145 Sulphur has obvious advantages, but it can be removed by hydrogenation. As a result, permanent additives that could limit the ensemble size (Ir, Re, etc.) are being explored. In catalytic reforming, studies are being carried out to determine the importance of ensemble size control in bi-functional and multi-functional catalysts [27]. The extension of the idea to other systemsincluding non-metallic acid oxide catalysts-is also under consideration. Control of nucleation of coke on a catalyst is difficult since, in order to keep the surface area high, small particles are desired. Almost by definition, this means that a degree of crystalline disorder will result and that it will be impossible to control nucleation sites. There are more opportunities with bulk metals, and these are discussed in the section below. The other major route by which the formation of coke may be minimized involves the gasification of coke precursors or, with more difficulty, coke. There are several possible gasifying reactions, including C+2Hz-CH4
(1%
C+HzO-CO+H,
(16)
c + o* -
(23)
co2
c+coa-2co
(24)
Of these, the reaction with oxygen is important mainly during regeneration while the other reactions may be favoured during normal operation. The importance of these reactions depends on the thermodynamics of the system, the concentration of gasifying agent and the absence or presence of a catalyst. In some cases, the catalyst for the desired reaction also promotes gasification (e.g. nickel, iron, cobalt [28, 291) while in others a gasification catalyst is built into the the catalyst for the main reaction (Ir is added to Pt in a supported reforming catalyst [27]; alkali is added to a Ni/AlaOs steam reforming catalyst [30,31]). The thermodynamics of the system dictate operating conditions, which are adjusted to optimize the main reaction, minimize coking and maximize gasification. No general guidelines can be given. The first requirement for an adequate supply of gasifying agent involves efficient mass transfer, and this may be difficult in processes carried out with liquids. As a result, it is sometimes desirable to choose a solvent that can also act as a hydrogen carrier. During the hydroprocessing of very heavy oils, for example, tetralin is often chosen as a solvent as a result of the sequence of reactions [l] tetralin + RNHz --+ naphthalene
+ 2Hz -
naphthalene tetralin
+ RH + NH3
(25) (26)
the catalyst apparently being more effective for reaction (26) than for reaction (25). Tetralin is the classic hydrogen carrier for such systems, but there is evidence that other solvents may also be used [l ,2]. The second requirement can be discussed more easily by reference to gas-solid systems. The active
gasifying agent for coke precursors in such systems is usually not molecular hydrogen or steam, but species derived therefrom. Adsorbed atomic hydrogen or OH species are more reactive and gasification catalysts are known to promote the formation of such species as well as the reaction between coke and the adsorbed reactant. Thus, for example, Urania and magnesia are added to steam reforming catalysts [31], at least in part because they promote dissociative adsorption of steam. Iridium, added to reforming catalysts [27], promotes both dissociative adsorption of hydrogen and reaction of the hydrogen species formed with coke [32]. Further advantage may be gained if the adsorbed gasifying agent can migrate across the surface. Steam reforming is a good example, since the main reaction and coke formation proceeds on nickel, while dissociative adsorption of steam occurs on the support [30, 311. As a result, the active species must be able to migrate. The catalyst designer can choose components that favour mobility of adsorbed species, and this is often done. Thus, for example, iridium favours mobile adsorption of hydrogen [33], and is added to reforming catalysts [27]. Although the discussion above has concentrated on carbon gasification, coke precursors are usually easier to gasify. They are more reactive and often contain hydrogen themselves [S] . As a result gasification of coke intermediates is more facile. Similar arguments apply to non-metallic catalysts, but problems may arise with acidic oxides and sulphides. Most gasification catalysts produce atomic hydrogen (metals, metal oxides, metal sulphides) or oxygen containing entities (OH-, O*-). Regrettably, basic oxides are the best catalysts for the latter and these cannot be added to, for example, silica-alumina without reducing acidity and overall catalytic activity. As a result, coke gasification during the main reaction is not promoted. Gasification of coke precursors and coke is seen to depend on operating conditions (which keep the level of the gasifying agent high) and on catalyst design. There is little that an operator can do to control the latter, but catalyst selection may be made with these concepts in mind. 5. Minimization burization
of unwanted
effects
due to car-
The formation of coke during a reaction is, almost without exception, an unwanted phenomenon that cannot be eliminated completely. Provided the intervals between shut-down for coke removal are long, economic operation is possible. Some of the problems caused by coke are, however, more difficult to handle. Carburization of alloys used to construct the plant is one such problem, and attention is focused on this phenomenon in this section. The nature of the problem has been outlined above. If coke is deposited on a metal surface, there is a tendency for the carbon to dissolve in and migrate through the metal. Metal carbides may be formed, or carbon may
146 Studies of the carburization resistance of 21 alloys have been carried out under conditions pertinent to the steam cracking reaction (Table 2) [34]. Under pyrolysis conditions, alloys PG2535 Nb, G4868, GHR35CW and 36XT were found to be most resistant to carburization, while under steam cracking conditions, alloys PG2535Nb, 2535Sp, 2833Si, PG2535H and G4868 were found to resist carbon uptake most strongly. These comparative ratings are useful as they are, but a wider range of possibilities could exist if the reasons for their behaviour were understood. The role of silicon is reasonably clear [35,36]. Under steam cracking conditions, a layer of silica or silicate is produced on the surface and this protects the alloy against carburization. The scale may rupture with temperature fluctuations or metal creep, but protection will be provided while the silica layer is reasonably intact. The role of silica does not, however, explain the overall behaviour of the alloys, since relatively small amounts of strong carbide forming additives (e.g. Cr, Nb, Ti, Al) also have an effect on carburization resistance (compare preferred alloy compositions). The reasons for this are far from clear, and appear to differ with additive, but the benefits in extended tube life are significant. The finding that a silica containing layer may be formed on a surface leads directly to the suggestion that metal surfaces could be coated by an inert material to reduce coke formation and carburization. From the geometry of the systems, such coatings should be deposited from the vapour phase (e.g. sulphur from a gaseous sulphur containing molecule [24,37], silica from silicon tetraethyl [38]) where this is possible. Several examples of such effects have been quoted. Traces of sulphur have been found to reduce coke formation on alloys [39], apparently by a mechanism of
precipitate at a discontinuity in the ahoy. This, in itself, increases stress in the tube and, if the carbon is removed on regeneration and then replaced (by subsequent operation), stress, corrosion and metal dusting all increase. The net result can be catastrophic tube failure in the last analysis (Fig. 9). The process involves three stages and all may be controlled to some extent. Coke formation is followed by dissolution/migration and, finally, carburization. Control of coking has been dealt with above but control of dissolution/migration and of carburization has not. The first requirement is a smooth surface. Carbon formation is known to be affected by the crystal face of the metal exposed to the gas [ 17, 181 and by the roughness of the surface [ 171. Protrusions catalyse coking, appearing to act as nucleation sites for coke formation and as dissolution entry points. This latter effect may well result from local supersaturation of the metal caused by the higher surface area of the protuberance. Minimization of the effect may be.achieved by smoothing the surface which, in industrial terms, is normally carried out by machining the tube [ 151. Although an expensive operation, the subsequent gain in control of coking is sufficient to repay the cost. Perhaps the most important means of controlling dissolution/carburization is, however, by careful selection of materials or by surface coating. There is a constraint, because carburization occurs at high temperatures where a nickel based alloy is necessary, but, even within this, it is possible to reduce carburization by several factors. The rationale behind materials selection is best described for steam cracking. This is a pyrolytic reaction in which steam is used as a diluentlgasifying agent [15] : the steam is present in ratios such that chromium and silicon on the surface of the alloy will be present as oxides but iron and nickel exist as metals.
TABLE 2. Compositions of alloys. Alloy
Ni
Cr
Nb
W
Si
G4848
19.8
24.9
-
_
HK40
21.6
24.5
-
-
Mn
C
Other
1.8
1.1
0.39
_
1.3
0.6
0.4
0.8 MO
IN519
24.2
23.5
1.83
_
1.2
0.9
0.27
-
G4868 31H 800 800H 801 PG2833Si 36XS 36XA 36X PG2535H G4857 PG2535Nb KHR35CW 2535s~ 2535Mo 36XT PG2848S G4879
29.1 32 32 32 32 32.7 34.1 34.2 34.4 34.8 35.2 35.2 35.4 35.6 37.4 44.8 45.7 45.70
29.3 21 21 21 21 27.4 25.9 25.1 23.8 25.3 27.2 25.0 25.0 26.0 26.1 35.51 27.7 29.70
_ _ _ 1.11 0.8 0.73 1.57 1.40 _ 1.34 _
_ _ 1.52 _ 1.20 _ 1.70 4.25 4.45
2.2 0.6 1 1 1 2.4 1.6 1.5 1.3 2.4 1.6 2.2 1.6 1.8 1.4 1.6 1.8 1.5
1.5 0.6 1.5 1.5 1.5 0.9 1.2 1.4 1.2 1.0 0.7 1.0 0.9 0.3 1.2 0.9 1.1 0.6
0.51 0.1 _ _ 0.41 0.43 0.43 0.40 0.45 0.47 0.45 0.40 0.39 0.43 0.35 0.57 0.55
0.4 0.4 0.4 1.2
-
Ti, 0.4 Al Ti Ti Ti
3.3 Al
0.4 MO 0.2 MO 1.5 MO
147 ensemble size control (see above). Silica [38] and alumina [39], both of which can be deposited from the vapour phase, have been found to reduce coking and carburization. Phosphorus and bismuth, or bismuth alone, also reduce coking [40], but it is uncertain whether this results from interactions in the gas phase or on the surface, There Seems little doubt that surface coatings have considerable potential. If the coating is applied before operation, there may be a need for repeat coating after a decoking cycle, or there may be a need for continuous addition of the coating agent (e.g. sulphur), but the net gain in minimization of coking and carburization has considerable economic importance. Other coatings may also be of interest, and some contending materials are currently under study. From the above, it is clear that careful selection of materials of construction is necessary to avoid problems caused by carburization. There are benefits resulting from smoothing reactor surfaces and from applying inert surface coatings. The latter development is comparatively recent, and techniques may not be generally available at the moment.
TABLE 3. Melting points and temperatures at which of sintering can be expected for some catalysts
the onset
Catalysts
Melting point (“0
Metals
cu 11 Fe
1495 1083 2410 1535
Ni Pd Pt Re Rh Ru Ag Sn Zn
1453 1552 1772 3180 1966 2310 962 232 420
Al203 SiO.2
2015 1713 High but varying with composition 2800
900-1400
1975
650-1000
Oxides
Co
SiOz/A1203 Zeolites Mgo
I
ZnO
Sintering point (“Cl 500-750 360-540 800-1200 500-750 500-725 500-750 570-880 1000-1500 600-1000 730-1200 330-480 77-110 140-210 670-1000 570-850
6. Coke removal The accumulation of coke can be minimized but usually cannot be eliminated. As a result, it is necessary to stop operations from time to time in order to remove coke from the catalyst and the reactor. The periods between removal should be as long as possible and coke removal, when necessary, should be efficient and fast. Large deposits can be removed mechanically, but smaller deposits are generally removed by gasification with steam or oxygen: C+HzO-CO+H,
(16)
c + 02
(23)
-
co2
Depending on the nature of the surface on which coke is deposited, these reactions may be catalysed or not. Reaction (23) is highly exothermic and considerable care must be taken when regenerating with oxygen. Catalysts may be permanently deactivated by sintering (loss of surface area) and this is known to start to be significant at temperatures between 33% and 50% of the melting points of the solids [41]. Inspection of Table 3 shows that maximal temperatures can be faily low, and careful control of burn-off is necessary. This is usually achieved by a slow step-wise increase in the temperature and the amount of oxygen fed to the bed. Thus, for example, a reforming catalyst may be regenerated by heating at 475 “C (oxygen slowly increased from 1% to 5% over about 24 h) and then at 510 “C for 2 h under 21% oxygen [27]. At all times, the temperature is carefully controlled so as not to exceed the critical values for platinum or alumina. Oxygen is the preferred gasifying agent in many cases since steam can accelerate sintering [42]. The steamcarbon reaction is easier to control, but metal oxides, in particular, sinter more readily in steam. This may even
present problems during gasification with oxygen, since partially hydrogenated residues will produce steam, and efficient product removal is desirable. In other cases, however, the effect of steam is less and the sensitivity of the system to over-temperature is greater, leading to preferred gasification with steam. Raney copper based catalysts are a case in point [43]. Recovery of heat liberated during coke removal is one of the few positive factors about regeneration, and in some cases it can be critical to the whole operation. Catalytic cracking is a case in point, where cracking is endothermic and uses the heat generated by coke removal [2]. In this case, it is important to recover all of the heat, and partial combustion 2c+o*-2co
(25)
is undesirable. As a result, it has become recent practice to add a trace of platinum to the silica-alumina [2], the metal having little effect on cracking but promoting complete oxidation: 2co + 02 -
2co*
(26)
The removal of coke from reactors also poses problems, but for different reasons. The deposits are usually thicker and, as a result, the catalytic action of the underlying metal is minimal. This results in the necessity of gasifying coke at higher temperatures, and steam cracker tube regeneration is usually carried out at about 600 “C rising to 800 “C. The initial gasifying agent is steam, an eventual mixture of steam and air (-50% of each) being introduced slowly. In these cases, sintering is less of a problem but carburization and metal dusting must be prevented. Overtemperatures increase carburization, a process that should be limited where possible. Metal dusting may occur when coke deposits within the alloy are gasified,
148 leading to physical separation of metal particles from the tube. There may be a case for gasifying only that carbon deposited on the alloy (and not the carbon dissolved in the alloy), but this is open to question since it would leave the tube in a carburized state. Studies are currently in progress to determine optimal decoking/recoking procedures.
7. Conclusions The formation of coke on catalysts and reactors is an unwanted side process that originates from several sources. Reactions in the gas and liquid phase and on the surfaces of solids may produce coke, and this can deactivate catalysts, increase heat transfer, block reactors and tubes, and cause high temperature corrosion. Minimization of coking can be achieved by careful control of individual systems. The thermodynamics of the desired and coking reactions dictate optimal operating conditions. Gas phase coking can be prevented by increasing dilution and surface:volume ratios. Catalytic coke control largely involves the catalyst designer/ manufacturer, but the chemical engineer may choose suitable catalysts to minimize the effect. Carburization may be reduced by careful choice of materials and by coating the surfaces with an inert material. Coking cannot normally be eliminated. Coke removal may be achieved mechanically or by gasification with oxygen or steam. Care is necessary to avoid high temperatures during the latter process.
References 1 0. Weisser and S. Landa, St&hide Catalysts: Their Properties and Applications, Pergamon, Oxford, U.K., and Viewig, Brunswick, F.R.G., 1973. 2 B. C. Gates, J. R. Katzer and G. C. A. Schuit, Chemistry of Catalytic Processes, McGraw-Hill, New York, 1979. 3 B. H. Thompson and C. T. Brooks, 165th Am. Chem. Sot. Natl. Meet., Dallas, TX, 1973. 4 P. S. Virk, L. E. Chambers and H. N. Woebcke, Adv. Chem. Ser., I31 (1974) 237. 5 S. M. Davis, F. Zaera and G. A. Somojai, J. Catal., 77 (1982) 439. 6 D. L. Trimm and C. J. Turner, J. them. Tech. Biotechnol., 31 (1981) 285. 7 S. C. Graham, J. B. Homer and J. L. J. Rosenfeld, Froc. R. Sot., Ser. A, 344 (1975) 259. 8 S. C. Graham,fioc. R. Sot., Ser. A, 377 (1981) 119. 9 D. L. Trimm, Cstal. Rev. Set. Eng., 16 (1977) 155. 10 J. G. McCarty and H. Wise,J. Catal., 57 (1979) 406.
11 J. R. Rostrup-Nielsen and D. L. Trimm, J. CataL. 48 (1977) 1.55. 12 W. G. Appleby, J. W. Gibson and G. M. Good, Ind. Eng. Chem. Process Des. Dev., I (1962) 102. 13 J. B. Butt and R. M. Billamora, Chem. React. Eng. Rev., Am. Chem. Sot. (Houston), (1978) 288. 14 R. Pearce and W. R. Patterson, Catalysis and Chemical Processes, Blackie, London, U.K. 1981. 15 L. F. Albright, B. L. Crynes and W. H. Corcoran, Pyrolysis. Theory and Industrial Practice, Academic Press, New York, 1983. in S. A. Jansson and Z. A. Foroulis (eds.), 16 1. Koszman, High Temperature Gas-metal Reactions in Mixed Environments, Met. Sot., A.I.M.E., New York, 1973, p. 155. 17 C. A. Bernard0 and D. L. Trimm, Carbon, 14 (1976) 225. Catal, 62 (1980) 35. 18 B. J. Cooper and D. L. Trimm,J. 19 D. L. Trimm and C. J. Turner, J. Chem. Tech. Biotechnoi., 31 (1981) 195. 20 N. M. Laurendeau, Frog. Energy Combust. Sci.. 4 (1978) 221. 21 N. N. Semenov, Some Problems of Chemical Kinetics and Reactivity, Pergamon, Oxford, U.K., 1958. 22 G. C. Bond, Catalysts by Metals, Academic Press, New York, 1962. Catal., 5 (1983) 263. 23 D. L. Trimm,Appl. 24 J. R. Rostrup-Nielsen, J. (7ataZ.. 25 (1984) 31. 25 J. J. McCarroll, Surf. Sci, 53 (1975) 297. 26 D. Briggs, J. Dewing, A. G. Burden, R. Moyes and S. P. B. Wills, J. Catal., 65 (1980) 31. 27 D. A. Dowden, Catalysis, Spec. Period. Rep. Chem. See., 2 (1978) 1. 28 R. T. Rewick, P. R. Wentrcek and H. Wise, Fuel, 53 (1974) 274. 29 C. A. Bernard0 and D. L. Trimm, Carbon, 17 (1979) 115. 30 J. R. Rostrup-Nielsen, Steam Reforming, Teknisk Forlag, Copenhagen, Denmark, 1975. 31 J. R. H. Ross, Surface and defect properties of solids, Spec. Rep. R. Sot. Chem., 4 (1975) 34. 32 R. T. K. Baker and R. D. Sherwood, J. CataL, 61 (1980) 378. 33 S. J. Tauster and S. C. Fung, J. Catal.. 55 (1978) 29. 34 D. J. Young, G. M. Smith and D. L. Trimm, Proc. 35th Aust. Inst. Metals Conf., Sydney, Australia, 1982, p. 86. 35 A. Schnaas and H. J. Grabke, Oxid. Met., I2 (1978) 387. 36 K. Ledjeff, A. Rahmel and M. Schorr, Werhst. Korros., 31 (1980) 38. 37 A. HoJmen, 0. Lindvag and D. L. Trimm, J. Chem. Tech. Biotechnol., 31 (1981) 311. 38 W. Karcher and P. Glaude, Carbon, 9 (1971) 617. 39 R. H. Long, M. C. Sze and H. Under, Met. Prog. (1975) 52. 40 I. Kozman, U.S. Patents 3 546 317 and 3531394,197O. 41 D. L. T&m, Design of Industrial Catalysts, Elsevier, Amsterdam, 1980. 42 D. L. Trimm, D. J. Young and P. Udaja, m B. Delmon and G. F. Froment (eds.), Studies in Surface Science and Catalysis, 1980, p. 331. 43 W. L. Marsden, M. S. Wainwright and J. B. Friedrich, Znd. Eng. Chem. Prod. Res. Dev., 19 (1980) 551.
Control of Coking D. L. TRIMM School of Chemical Engineering and Industrial Kensington, N.S. W. 2033 (Australia)
Chemistry,
University
of New South Wales, P, 0. Box 1,
(Received August 5, 1983)
Abstract The formation of coke during the large-scale processing of hydrocarbons is largely unwelcome. Not only does it represent a loss of product but it also leads to decreased heat transfer and to blockage. At high temperatures, carburlzation of metallic materials can occur, leading to catastrophic failure. The origins of different types of coke are discussed, and coke deposition is shown to be a balance between coke formation and coke removal. Carburization can occur when coke is deposited on metal surfaces. Means of minimizing coking and carburlzation are considered in terms of the operating parameters open to control and in terms of choice of materials. If coke does accumulate, the means of coke removal are presented, with particular attention being focused on the need to control carefully the conditions of decoking. 1. Introduction The industrial processing of organic feedstocks to produce desired chemicals involves a variety of reactions carried out over a wide range of conditions. The reactions may be catalytic or noncatalytic, may be in the liquid phase or the gas phase, and may be carried out at high or low temperature or pressure. Each system presents individual challenges to the chemist and chemical engineer but some factors are common to most. Of these, coking is of major importance since it often dictates the overall economics of the process. Coking is a collective noun, describing the accumulation of carbonaceous residues in the reactor or in the plant. The nature of the residue varies, from polymeric deposits favoured at low temperatures to carbons produced under more extreme conditions. The residues deposit on different surfaces throughout the plant, leading to deactivation of catalysts, mass transfer problems, decreased heat transfer and increased pressure drop. They may be removed mechanically or by gasification, but this entails interruption of normal plant operation and may involve degradation of the catalyst or the reactor material. As a result, there is a strong incentive to minimize the problem and to optimize conditions for removal. Removal of coke is a reasonably straightforward procedure, with care being necessary to avoid damage to a catalyst by, for example, overheating. Minimization of coke formation requires more detailed knowledge of the origins of coke since these can vary markedly. In general, coke formation is a polymerization/dehydrogenation process which is favoured by high temperatures and low pressures; in particular, the chemical reactions producing
coke can be very different. As a result, it is convenient to begin this article by considering the nature and origins of different types of coke. It should be remembered that coking is an undesired side reaction occurring concurrently with a desired main reaction. As such, process conditions are designed to minimize the former and to maximize the latter. However, it is the conditions appropriate to the desired reaction which are more important since these create boundary conditions within which coking should be minimized. Since coke is formed in many different processes, the operating conditions involved also vary widely. As a result, it is not surprising that both the nature and the origin of coke vary markedly. In this article, attention is focused first on the different types of coke and where they come from. The problems that they can cause are then discussed. Control of coking is considered and, if this falls, removal of the accumulated deposit and the problems arising during this are discussed. 2. Nature and origins of coke 2.1. Low temperature
liquid phase systems
Formation of a polymeric residue which has many similarities to coke can be observed at quite low temperatures in some systems. Thus, for example, liquid phase hydrogenation of oils can lead to deactivation of the catalyst by coking [I, 21, and pipeline transfers of hot oils can lead to accumulation of residues in regions of stagnant flow [2]. The rates of residue formation are low, but the deposit appears to originate from polymer-
Gem. Eng. F’roces.. 18 (1984) 137-148
0 Elsevier Sequoia/Printed
in The Netherlands
138 Polymerization can be minimized by adding a free radical capping agent such as supplementary hydrogen [3,41> .
ization of olefinic, naphthalenic and aromatic compounds, followed by slow dehydrogenation and rearrangement reactions to form stable coke deposits. The reactions involved, being slow, are open to question, but it is believed that the general type of process involved can be represented by the sequences below [3,4]. For clarity, the feedstock is considered to be butadiene.
<
+
a,- a
oj
+ 0
(5)
but the stability of some of the larger free radicals makes this process less efficient than would be expected. As a result, coke formation can be reduced but not eliminated. 2.2. Gas phase coke
+2Hz
(2)
+ H’
(3)
.
+ CzH4
+Hz-2
2
-
Q
-
+ H’
(4)
3 co Free radicals are believed to be important in the processes and, at low temperatures, resonance stabilized free radicals based on aromatic compounds
can be relatively long lived [3,4]. Their concentration is low, but their long lifetimes increase the chance of molecular weight build-up by polymerization, particularly in stagnant zones in the reactor or pipeline. These slow polymerization reactions increase molecular weight and decrease solubility, causing precipitation on any convenient surface. Once precipitated, the compounds are not necessarily inert, and further liquidsolid free radical polymerizations can occur [5] leading to growth of the deposit. The material can also rearrange to form a more ordered structure (usually involving elimination of hydrogen) but this is slow at low temperatures.
Coke formation at higher temperatures is more important and can occur in the gas phase or on a surface. The importance of either depends on the reactivity of the coke producing gas and the nature of the surface but, in general, gase phase coking is more important at higher temperatures (L 600 “C). Gas phase carbons are believed to be produced by polymerization/dehydrogenation processes, and the deposits are known to contain a range of polycyclic aromatic materials together with carbon. The carbon (gas phase carbon) is generally disordered (Table 1) and has the appearance shown in Fig. 1. The polycyclic aromatics (tars) are difficult to distinguish from carbon since they have the same appearance and have a high carbon:hydrogen ratio. They can be dissolved in suitable solvents for analysis, and this shows that components ranging from naphthalene (two aromatic rings) to hexabenzotriphenylene (ten aromatic rings) are present. Calculations of the vapour pressures of these compounds show that they can exist in the solid phase, even at temperatures up to 1000 “C, and, as a result, will behave much as carbon itself. There seems little doubt that tars and carbon are produced by free radical reactions similar to those described above. Studies of gas phase reactions show that, with increasing residence time, gas phase products, tars and gas phase carbon are produced in turn and that tars and carbons have a well-defined induction period (Fig. 2) [6]. As a result, it is logical to suggest that the feedstock produces monomers that polymerise to greater extents over longer times, eventually forming first tars and then carbon. Undoubtedly this does happen, but it is not the complete explanation, since tars and carbon can be formed at exactly the same residence time in some systems [7, 81. We shah see that
TABLE 1. Typical properties of carbons
Spe
Typical formation temperature CC)
C:H ratio
Lc (A)
LA (A)
d (A)
Gas phase Surface Catalytic Qaminar) Graphite
1000 1000 600 2000
8 >80 20
12.2 28 200
42 51 400 OD
3.6-3.7 3.45-3.55 3.4 3.35
The data refer to crystals in which Lc is the length and LA the breadth. The crystals contain layers of carbon atoms parallel to each other at distance d apart.
Fig. 1. Photographs of typical coke.
volved are mainly dehydrogenation and ordering, the latter process involving a reorganization of randomly deposited coke to give, as a final product, an ordered graphite-like structure. Complete ordering rarely occurs and the carbon may, or may not, have a distinctive appearance (Fig. 1). If the original deposit is gas phase carbon, a liiited amount of ordering does not lead to a major change in morphology (Fig. 1); if the original deposit is tar, the surface carbon will have a higher degree of long-range order. 2.3. Cattilyticcoke
Fig. 2. The production of material during steam cracking as a function of residence time.
the induction period for tar and carbon production has important ramifications for other aspects of coking. Tars and gas phase carbon will collect on any convenient surface and, even if the surface is non-catalytic, further reactions will occur to produce a more ordered deposit of surface carbon (Table 1). The reactions in-
Coking is also affected by the presence of a catalyst such as a transition or noble metal (catalytic carbon). Highly crystalline deposits of carbon can be formed at relatively low temperatures (-400 “C) and no tars are produced. In addition to metal carbides, three types of carbon are seen-laminar graphite, non-oriented carbon and whisker carbon (Fig. 3). Laminar graphite tends to form under conditions of high temperature and low hydrocarbon pressure, while non-oriented carbon is favoured under conditions that favour high supersaturation of carbon in the metal. Carbon whiskers are produced mainly on transition metals and can be up to 7 pm long and 0.1 w wide.
ably well understood, but they may be complicated by mass and heat transfer effects (in reactions at high pressure, for example.) Coking, in these systems, results mainly from acid catalysed polymerization reactions of carbonium ions [ 121. Reactions such as cracking involve the formation of carbonium ions from olefins: RHCH=CHR’H + H+A- e RHCH,-C+HR’H S H,;R’H
t A- e
RHCH,-&HR’H
t A-
R=CH2 + HzC+R’H
(6) (7)
(8)
H+A- + R’=CH?
The carbonium ions may react with other olefins or with aromatics to give higher molecular weight species and, eventually, carbonaceous deposits: RCH,-:HR’
t RCH=CHR’ e
RCH2-FHR’
(9)
R&H-CHR’ t Fig. 3. Catalytic carbons: A, whisker carbon; B, non-oriented
carbon; C, graphite carbon.
RCH,--dHR’
The production of catalytic carbon can best be described with the aid of Fig. 4 [9]. Hydrocarbons adsorb on a metal surface and may react to produce gas phase products or dehydrogenated adsorbed intermediates. This process continues until carbon is produced on the surface which, in turn, can isomerize to other forms of carbon [5,6]. Depending on the metal, up to four forms of carbon have been identified ranging from o-carbon (very reactive), &carbon (amorphous and less reactive) and crystalline carbon to metal carbides of varying stoichiometry [IO]. The subsequent fate of this carbon determines the nature and amount of coke formed. o-carbon may be gasified or may react to produce P-carbon. &carbon or crystalline carbon may be gasified, but are more likely to encapsulate the surface, leading to laminar graphite or non-oriented carbon. Alternatively they may dissolve in the metal, diffuse to a grain boundary and precipitate out, producing metal carbides and whiskers of carbon with metal particles at their tips [9, 111. This process weakens the metal (see later). The relative amounts of different types of carbon then depend on the production of the intermediate (which can be reduced by efficient gasification), on the balance between the different forms of carbon and on the balance between carbon dissolution and encapsulation. Coke may also be formed on non-metallic catalysts and coking of acidic oxides and sulphides is well established. The underlying reaction mechanisms are reasonHydrocarbon
tl Dissolved carbon
Fig. 4. Model for catalytic carbon production.
Dissolved carbon
+0
+ A- ~~~H-H~~H+A_
(10) Dehydrogenation and isomerization will then occur but the resulting coke is less well ordered in its final form as a result of the mechanism of formation. Coke formation on sulphides appears to result from several causes and is of particular interest because these catalysts are often used to hydrotreat heavy oils [l, 21. The high molecular weight of the feedstock favours coke formation and excess hydrogen is used in an attempt to reduce the problem [ 11. The mechanism of coking may involve ionic and non-ionic intermediates, and it is difficult to tell which process may be more significant. Mass transfer processes may also influence coking, since hydroprocessing of heavy oils is often a gasliquid-solid reaction. Transfer of hydrogen through the liquid may be too slow to stop coking and intermediates may be trapped in pores to produce a local concentration excess that favours coking. It is seen that the formation of coke is a complex process that can involve many different processes. Before considering how these may be minimized, it is useful to recognize what kinds of problem can arise as a result of coke formation. 3. Problems caused by coking Most industrial chemical processes involve a catalyst, and it is not surprising that catalyst deactivation as a result of coke formation is perceived to be a major problem. The effect of different forms of coke on the kinetics of complex reactions has received much attention [ 131 and the deactivation/regeneration cycle is of major importance in determining the overall economics of a process [14]. There is little doubt that minimizing catalyst coking is a matter of considerable interest.
141 Coking is of major interest to chemical engineers, however, for other reasons, and these may be illustrated by reference to a flow diagram of a steam cracking plant (Figs. 5 and 6) [ 151. The process involves pyrolysis of a light feedstock (usually naphtha, ethane or propane) in the presence of steam to produce light olefins and other valuable products. The feedstock is mixed with steam and passed to a furnace countercurrent to the flue gas flow (Fig. 5). The gases pass through a convection heating section and a radiant coil before leaving the furnace to a transfer line heat exchanger, a water quench and a separating train (Fig. 6). The outputs from two furnaces are often combined in a bobbin before passage to the transfer line exchanger (TLE). Typical operating conditions involve a residence time in the radiant zone of 0.2-0.55 s at a pressure of 0.150.2 X lo6 Pa. The steam-hydrocarbon ratio in the feed depends on the nature of the hydrocarbons, varying from 0.25-0.4 (weight basis) for ethane to 0.8-1.0 for gas oil. The temperatures in the convection and radiant zones are about 600 and 900 ‘C respectively, and the specific heat requirements are high (-5000 kJ/kg for
naphtha-steam). As a result, high heat transfer rates are required. The gases that leave the furnace contain significant quantities of heat and much of this can be recovered by use of a TLE in which the gas temperature is dropped from around 900 “C to about 450 ‘C. The gases themselves contain a mixture of valuable components which must be separated, dried and purified. A typical processing train is shown in Fig. 6. After cooling in the TLE, residual free radicals are deactivated and some unwanted products removed in the oil quench. Heavy ends are then separated by fractionation and acid gases removed from the light ends. After drying, subsequent separation gives C1, CZ, C3 and C4 streams together with residual naphtha that can be used as gasoline. Acetylenic products are a problem in the C&C,, streams and must be removed, usually by selective hydrogenation. Separation of the remaining C2 and C3 products yields the major products, ethylene and propylene, the corresponding paraffins being recycled to the cracker. This type of plant layout is typical of many processes, with the exception that the temperature is high (-900 ‘C in the radiant zone) and that there is no catalyst. It is a convenient reference, however, to illustrate other problems caused by coke formation. The first of these involves materials of construction. The length and diameter of the reactor in the radiant zone are 90-120 m and 9-10 cm respectively. Because of the high heat transfer requirement, the tubes are made from heat resistant nickel-chromium alloys and coking catalysed by these alloys has been found to occur. The temperatures in the radiant zone are about 900 ‘C, and gas phase coking produces tars, gas phase carbon and, after deposition and rearrangement, surface carbon. Tars may deposit in the reactor or in the
TRINSFER UNE HEAT EXL”Plmjl
Fig. 5, Reactor/furnace arrangements for the steam craker.
ETH4NE OR PROPANE OR NAPHTHA OR ^.^ ^. STEAM CRACKER FUEL OIL I
ETHANEIPROPANE
I I
“2 RESIDUE
CZH2
REMOYAL
ETHYLENE FRACTION
GAS
ETHYLENE
I
PROPYLENE I
) Fig. 6. Typical processing train for steam cracking.
PYROLYSIS GASOLINE
142
bobbin/TLE, where a sharp drop in temperature occurs. Because of the induction period for the production of tars and carbon (see above) the convection zone and the early part of the radiant zone of the tube remains relatively coke free, but the back end of the tube, the bobbin and the TLE coke heavily. A typical deposit on a tube is shown in Fig. 7, and it is obvious that the thick layer of coke will increase significantly the heat transfer across the tube. As a result, the second problem arising from coking is due to decreased thermal efficiency. Coke deposited on a tube is not inactive, but diffuses, in time, into the tube material. This occurs preferentially at grain boundaries in the metal, leading to carburization (Fig. 8) and to higher thZr normal stress in the noncarburized parts of the tube. Creep often becomes critical, particularly as the metal temperature may be close to 1100 ‘C (to maintain the gas temperature at around 900 “C). Carburization of the alloys, then, is the third problem caused by coke. Once a thick deposit of coke has accumulated, it is necessary to decoke the tube by gasification with a steam-air mixture. This involves taking the tube out of service and cycling the temperature of the tube to achieve even burn-off. The expansion/contraction of the tube, together with gasification of carbon in the grain
Fig. 8. Carbon penetration tubes.
Fig. 7. Coke deposits on industrial steam cracker reactor tubes (a) and transfer line heat exchanger tubes (b). (Photographs courtesy of ICI Australia Pty. Ltd.)
,
of alloys used in steam cracking
boundaries, can lead to extensive metal dusting [16] (metal drops from the surface of the tube) and to production of a disrupted surface of composition differrent from that present originally. Subsequent re-coking will be accelerated by the surface roughness [ 17,181 and, possibly, by removal of the original layer of protective material on the surface (see below). This problem can be so acute that the life of a tube may be affected more by the number of decoking-recoking cycles than by extended operation under steady state conditions. Problems in the bobbin/TLE area are different but are also very significant. Although carburization can occur, tar condensation and reorganization is more important since this may lead to very bulky deposits and to physical blocking of the TLE tubes (Fig. 7(b)). The net effect of these problems is seen not only in decreased thermal efficiency or in metal dusting but also in catastrophic failure of tubes. Photographs of some failures observed during plant operation are shown in Fig. 9. In general, it is seen that coke formation can occur in the liquid or the gas phase, in the absence or presence of a catalyst, and at high and (relatively) low temperatures. It causes problems as a result of catalyst deactivation, of increased heat transfer resistance, of carburization/corrosion of reactor tubes and of physical blocking of equipment. As a result, there is considerable interest in minimizing coke formation: how this may be done is the subject of the text below.
143 from a series of free radical reactions that can be represented as [3,4, 191 R-R’
xR’ -
-
R’ t R”
(11)
coke
(12)
High temperatures and high pressures favour these types of reaction and the first essential is to keep temperature and pressure as low as possible. If this cannot be done, residence time in the gas phase at high temperatures should be minimized. This requires careful design not only of the reactor but also of preheaters, riser space and heat exchangers after the bed. The latter is particularly important in view of the fact that a known induction period exists for the production of tars and coke (Fig. 2). A second route to minimizing coking involves dilution. Coking is a polymerization reaction and, if the free radicals can be kept apart, coking will be minimized. As a result, there are advantages in using an inert diluent (nitrogen) or a reactive diluent (hydrogen or steam). The action of these last two gases also involves free radical capping. If reactions such as
(b)
Fig. 9. Industrial steam cracker tube failures. courtesy of ICI Austrialia Pty. Ltd.)
4. Control
(Photographs
R’+H,-RHtH’
(13)
R’-+HzO-RH+OH’
(14)
occur, the coke forming intermediates are replaced by low molecular weight free radicals, whose interaction will not form coke. In addition, gasification of coke or coke precursors may occur; this may be represented as C+2Hz-CH4
(15)
C+HzO-COtH,
(16)
of coking
Coking results from an accumulation of carbonaceous deposits from the gas phase or on a catalytic surface. In the present context, it is useful to consider minimization of gas phase coke formation, minimization of catalytic coke formation and minimization of deleterious effects due to coking. Since it is usually impossible to avoid coking altogether, a final section deals with coke removal. It is important to remember that the accumulation of coke results from a balance between coke deposition and coke (or coke precursors) removal. Control of coking then depends on minimizing deposition and maximizing gasification-either of coke or its precursors. This is distinct from coke removal during regeneration, which is a separate process involving interruption of normal operations. It is also necessary to remember that coking is an unwanted side reaction that occurs concurrently with a desired main .reaction. Any measures designed to minimize coking must take into account the operating conditions required for the desired reaction. ‘4.1. Control of coke from the gas phase Gas phase coking is dependent primarily on the nature of the gas and the operating conditions. It results
Reactions (13) and (14) can be considered as free radical deactivating reactions, and similar effects may be achieved in other ways. Thus, for example, free radical chains may be terminated at a surface [20] H’+H’+M-H2
tM
(17)
and, if the free radical hits a surface before it builds ulr size, the coking sequence can be interrupted before coke is formed.-As a result, there is an incentive to increase the surface:volume ratio in order to minimize gas phase free radical reactions. Of course, it is necessary that the pressure drop should not be too large and that the surface introduced should not be catalytic, but this may be achieved with relative ease. Similar effects can be achieved with gas phase additives, but these are rarely used in practice. Free radical termination reactions require removal of energy, either at a surface or to a molecule that can adsorb the energy. Fundamental studies use, for example, sulphur hexafluoride [20,21], but these materials are impracticable in an industrial situation. There is evidence that sulphur containing molecules present in a typical feed may act in the same way, but this is far from proven.
144 Control of gas phase coking then involves careful control of temperature and pressure, increasing dilution with an inert or a gasifying/free radical capping gas and optimizing surface:volume ratios in the heated space. 4.2. Control of catalytic coking The mechanism of coke formation on catalytic surfaces depends on the nature of the feed, of the catalyst and on operating conditions. Although the mechanism may differ, it turns out that means of controlling coke are very similar. In the first instance, these may be discussed with reference to the coking of metals. The process of coking is summarized in Fig. 4. Hydrocarbons adsorb on a surface to produce one or more intermediates. These may react to form the desired products or to form coke. The latter reaction, in the case of platinum, for example, involves the migration of an intermediate across the surface to a nucleation site where polymerization occurs to form a partially hydrogenated deposit [S]. This then reacts further, by dehydrogenation/isomerization, to form coke. On some metals (nickel, iron, cobalt, etc.) the coke may then remain on the surface or may dissolve in the metal 19,111. On the basis of this sequence, coke could be controlled by interference in several steps: (a) Temperature and pressure could be controlled so that the thermodynamic driving force for coking is minimal. (b) The intermediates on the surface could be controlled so that those involved in the desired reaction are maximized and those involved in coking are minimized. (c) Coke forming intermediates could be deactivated, for example by gasification. (d) Migration of coke precursors across the surface could be restricted. (e) Polymerization/nucleation could be minimized. (f) The partially hydrogenated deposits could be gasified rather than allowed to dehydrogenate. (g) Dissolution in the metal could be minimized. This is found to be more relevant to the following section, and is considered there. Control of temperature and pressure depends on the thermodynamics of the desired reaction, and is particular to a system. For any process, temperature-in particular-should be minimal although, as a catalyst deactivates, temperature is often increased to maintain production. This will, in turn, increase coking and lead, eventually, to complete deactivation. It may or may not be possible to control intermediates on the surface since a common intermediate may be involved in both the desired reaction and in coking. Methanation on nickel is a case in point [lo]. Carbon monoxide adsorbs to produce o-carbon, P-carbon, crystalline carbon and nickel carbide. o-carbon may react with hydrogen to form methane or may isomerize to form P-carbon. p-carbon is less reactive and rearrangement to crystalline carbon and coke is favoured. As a result it is impossible to control the nature of the intermediates-what must be done is to control the further reactions of o-carbon.
In other systems, it may be possible to control the intermediate. Ethylene, for example, may adsorb as a pi-bonded species or as a sigma-bonded species [22], with hydrogenation being favoured by the former and coke formation by the latter. In this case, it is obvious that we should choose a catalyst or adjust the catalyst to favour pi-bonded intermediates. Exactly how this is done is not easy to explain, since catalyst designers do not always understand what is going on with a particular catalyst. In some cases, however, it is possible to design a catalyst in this way. Coking involves migration over the surface and polymerization, and it has been found to be possible to control these processes by controlling the geometry of the surface. The principle of the method involves limiting the number of near-neighbour catalytically active sites (ensemble size [23]). Using methanation as an example, the desired reaction may be written in a simplified form as co-c+0
xx xx
(18)
Hz-2; 0+2Hx-H,0+4X xx &‘45f
-CH,+6X
while coking results from reactions such as 2xcx-
c-c xx xx
(22)
It is obvious that the ensemble size for coking must be much larger than that required for the main reaction and, as a result, it could be possible to limit the ensemble size to the point where the desired reaction proceeds and coking does not. At least in the case of nickel, this has been found to be possible by the addition of small amounts of a sulphur containing gas to the feed [24] _Too much sulphur poisons the catalyst, but a trace of sulphur adsorbs on nickel in a regular array which depends on the crystal face involved and the concentration of sulphur [25]. It turns out that the number of sites left uncovered is sufficient to allow the desired reaction to proceed, while not allowing coke formation. As a result, pilot plants have been run with small traces of sulphur in the feed to produce excellent results over long periods in the absence of coking [24]. The role of adsorbed sulphur may also be to reduce surface migration. Coke precursors migrate across a surface either by a desorption/re-adsorption mechanism or by movement in the adsorbed state [26]. In the latter case, the adsorbed species is loosely bonded to the surface and, if the nature of the surface changes (for example, at the boundaries of the ensemble where nickel is replaced by nickel sulphlde), then migration may be limited. The possibility of controlling coke by controlling ensemble size has been recognised only very recently [23,24], and the ramifications are still being explored.
145 Sulphur has obvious advantages, but it can be removed by hydrogenation. As a result, permanent additives that could limit the ensemble size (Ir, Re, etc.) are being explored. In catalytic reforming, studies are being carried out to determine the importance of ensemble size control in bi-functional and multi-functional catalysts [27]. The extension of the idea to other systemsincluding non-metallic acid oxide catalysts-is also under consideration. Control of nucleation of coke on a catalyst is difficult since, in order to keep the surface area high, small particles are desired. Almost by definition, this means that a degree of crystalline disorder will result and that it will be impossible to control nucleation sites. There are more opportunities with bulk metals, and these are discussed in the section below. The other major route by which the formation of coke may be minimized involves the gasification of coke precursors or, with more difficulty, coke. There are several possible gasifying reactions, including C+2Hz-CH4
(1%
C+HzO-CO+H,
(16)
c + o* -
(23)
co2
c+coa-2co
(24)
Of these, the reaction with oxygen is important mainly during regeneration while the other reactions may be favoured during normal operation. The importance of these reactions depends on the thermodynamics of the system, the concentration of gasifying agent and the absence or presence of a catalyst. In some cases, the catalyst for the desired reaction also promotes gasification (e.g. nickel, iron, cobalt [28, 291) while in others a gasification catalyst is built into the the catalyst for the main reaction (Ir is added to Pt in a supported reforming catalyst [27]; alkali is added to a Ni/AlaOs steam reforming catalyst [30,31]). The thermodynamics of the system dictate operating conditions, which are adjusted to optimize the main reaction, minimize coking and maximize gasification. No general guidelines can be given. The first requirement for an adequate supply of gasifying agent involves efficient mass transfer, and this may be difficult in processes carried out with liquids. As a result, it is sometimes desirable to choose a solvent that can also act as a hydrogen carrier. During the hydroprocessing of very heavy oils, for example, tetralin is often chosen as a solvent as a result of the sequence of reactions [l] tetralin + RNHz --+ naphthalene
+ 2Hz -
naphthalene tetralin
+ RH + NH3
(25) (26)
the catalyst apparently being more effective for reaction (26) than for reaction (25). Tetralin is the classic hydrogen carrier for such systems, but there is evidence that other solvents may also be used [l ,2]. The second requirement can be discussed more easily by reference to gas-solid systems. The active
gasifying agent for coke precursors in such systems is usually not molecular hydrogen or steam, but species derived therefrom. Adsorbed atomic hydrogen or OH species are more reactive and gasification catalysts are known to promote the formation of such species as well as the reaction between coke and the adsorbed reactant. Thus, for example, Urania and magnesia are added to steam reforming catalysts [31], at least in part because they promote dissociative adsorption of steam. Iridium, added to reforming catalysts [27], promotes both dissociative adsorption of hydrogen and reaction of the hydrogen species formed with coke [32]. Further advantage may be gained if the adsorbed gasifying agent can migrate across the surface. Steam reforming is a good example, since the main reaction and coke formation proceeds on nickel, while dissociative adsorption of steam occurs on the support [30, 311. As a result, the active species must be able to migrate. The catalyst designer can choose components that favour mobility of adsorbed species, and this is often done. Thus, for example, iridium favours mobile adsorption of hydrogen [33], and is added to reforming catalysts [27]. Although the discussion above has concentrated on carbon gasification, coke precursors are usually easier to gasify. They are more reactive and often contain hydrogen themselves [S] . As a result gasification of coke intermediates is more facile. Similar arguments apply to non-metallic catalysts, but problems may arise with acidic oxides and sulphides. Most gasification catalysts produce atomic hydrogen (metals, metal oxides, metal sulphides) or oxygen containing entities (OH-, O*-). Regrettably, basic oxides are the best catalysts for the latter and these cannot be added to, for example, silica-alumina without reducing acidity and overall catalytic activity. As a result, coke gasification during the main reaction is not promoted. Gasification of coke precursors and coke is seen to depend on operating conditions (which keep the level of the gasifying agent high) and on catalyst design. There is little that an operator can do to control the latter, but catalyst selection may be made with these concepts in mind. 5. Minimization burization
of unwanted
effects
due to car-
The formation of coke during a reaction is, almost without exception, an unwanted phenomenon that cannot be eliminated completely. Provided the intervals between shut-down for coke removal are long, economic operation is possible. Some of the problems caused by coke are, however, more difficult to handle. Carburization of alloys used to construct the plant is one such problem, and attention is focused on this phenomenon in this section. The nature of the problem has been outlined above. If coke is deposited on a metal surface, there is a tendency for the carbon to dissolve in and migrate through the metal. Metal carbides may be formed, or carbon may
146 Studies of the carburization resistance of 21 alloys have been carried out under conditions pertinent to the steam cracking reaction (Table 2) [34]. Under pyrolysis conditions, alloys PG2535 Nb, G4868, GHR35CW and 36XT were found to be most resistant to carburization, while under steam cracking conditions, alloys PG2535Nb, 2535Sp, 2833Si, PG2535H and G4868 were found to resist carbon uptake most strongly. These comparative ratings are useful as they are, but a wider range of possibilities could exist if the reasons for their behaviour were understood. The role of silicon is reasonably clear [35,36]. Under steam cracking conditions, a layer of silica or silicate is produced on the surface and this protects the alloy against carburization. The scale may rupture with temperature fluctuations or metal creep, but protection will be provided while the silica layer is reasonably intact. The role of silica does not, however, explain the overall behaviour of the alloys, since relatively small amounts of strong carbide forming additives (e.g. Cr, Nb, Ti, Al) also have an effect on carburization resistance (compare preferred alloy compositions). The reasons for this are far from clear, and appear to differ with additive, but the benefits in extended tube life are significant. The finding that a silica containing layer may be formed on a surface leads directly to the suggestion that metal surfaces could be coated by an inert material to reduce coke formation and carburization. From the geometry of the systems, such coatings should be deposited from the vapour phase (e.g. sulphur from a gaseous sulphur containing molecule [24,37], silica from silicon tetraethyl [38]) where this is possible. Several examples of such effects have been quoted. Traces of sulphur have been found to reduce coke formation on alloys [39], apparently by a mechanism of
precipitate at a discontinuity in the ahoy. This, in itself, increases stress in the tube and, if the carbon is removed on regeneration and then replaced (by subsequent operation), stress, corrosion and metal dusting all increase. The net result can be catastrophic tube failure in the last analysis (Fig. 9). The process involves three stages and all may be controlled to some extent. Coke formation is followed by dissolution/migration and, finally, carburization. Control of coking has been dealt with above but control of dissolution/migration and of carburization has not. The first requirement is a smooth surface. Carbon formation is known to be affected by the crystal face of the metal exposed to the gas [ 17, 181 and by the roughness of the surface [ 171. Protrusions catalyse coking, appearing to act as nucleation sites for coke formation and as dissolution entry points. This latter effect may well result from local supersaturation of the metal caused by the higher surface area of the protuberance. Minimization of the effect may be.achieved by smoothing the surface which, in industrial terms, is normally carried out by machining the tube [ 151. Although an expensive operation, the subsequent gain in control of coking is sufficient to repay the cost. Perhaps the most important means of controlling dissolution/carburization is, however, by careful selection of materials or by surface coating. There is a constraint, because carburization occurs at high temperatures where a nickel based alloy is necessary, but, even within this, it is possible to reduce carburization by several factors. The rationale behind materials selection is best described for steam cracking. This is a pyrolytic reaction in which steam is used as a diluentlgasifying agent [15] : the steam is present in ratios such that chromium and silicon on the surface of the alloy will be present as oxides but iron and nickel exist as metals.
TABLE 2. Compositions of alloys. Alloy
Ni
Cr
Nb
W
Si
G4848
19.8
24.9
-
_
HK40
21.6
24.5
-
-
Mn
C
Other
1.8
1.1
0.39
_
1.3
0.6
0.4
0.8 MO
IN519
24.2
23.5
1.83
_
1.2
0.9
0.27
-
G4868 31H 800 800H 801 PG2833Si 36XS 36XA 36X PG2535H G4857 PG2535Nb KHR35CW 2535s~ 2535Mo 36XT PG2848S G4879
29.1 32 32 32 32 32.7 34.1 34.2 34.4 34.8 35.2 35.2 35.4 35.6 37.4 44.8 45.7 45.70
29.3 21 21 21 21 27.4 25.9 25.1 23.8 25.3 27.2 25.0 25.0 26.0 26.1 35.51 27.7 29.70
_ _ _ 1.11 0.8 0.73 1.57 1.40 _ 1.34 _
_ _ 1.52 _ 1.20 _ 1.70 4.25 4.45
2.2 0.6 1 1 1 2.4 1.6 1.5 1.3 2.4 1.6 2.2 1.6 1.8 1.4 1.6 1.8 1.5
1.5 0.6 1.5 1.5 1.5 0.9 1.2 1.4 1.2 1.0 0.7 1.0 0.9 0.3 1.2 0.9 1.1 0.6
0.51 0.1 _ _ 0.41 0.43 0.43 0.40 0.45 0.47 0.45 0.40 0.39 0.43 0.35 0.57 0.55
0.4 0.4 0.4 1.2
-
Ti, 0.4 Al Ti Ti Ti
3.3 Al
0.4 MO 0.2 MO 1.5 MO
147 ensemble size control (see above). Silica [38] and alumina [39], both of which can be deposited from the vapour phase, have been found to reduce coking and carburization. Phosphorus and bismuth, or bismuth alone, also reduce coking [40], but it is uncertain whether this results from interactions in the gas phase or on the surface, There Seems little doubt that surface coatings have considerable potential. If the coating is applied before operation, there may be a need for repeat coating after a decoking cycle, or there may be a need for continuous addition of the coating agent (e.g. sulphur), but the net gain in minimization of coking and carburization has considerable economic importance. Other coatings may also be of interest, and some contending materials are currently under study. From the above, it is clear that careful selection of materials of construction is necessary to avoid problems caused by carburization. There are benefits resulting from smoothing reactor surfaces and from applying inert surface coatings. The latter development is comparatively recent, and techniques may not be generally available at the moment.
TABLE 3. Melting points and temperatures at which of sintering can be expected for some catalysts
the onset
Catalysts
Melting point (“0
Metals
cu 11 Fe
1495 1083 2410 1535
Ni Pd Pt Re Rh Ru Ag Sn Zn
1453 1552 1772 3180 1966 2310 962 232 420
Al203 SiO.2
2015 1713 High but varying with composition 2800
900-1400
1975
650-1000
Oxides
Co
SiOz/A1203 Zeolites Mgo
I
ZnO
Sintering point (“Cl 500-750 360-540 800-1200 500-750 500-725 500-750 570-880 1000-1500 600-1000 730-1200 330-480 77-110 140-210 670-1000 570-850
6. Coke removal The accumulation of coke can be minimized but usually cannot be eliminated. As a result, it is necessary to stop operations from time to time in order to remove coke from the catalyst and the reactor. The periods between removal should be as long as possible and coke removal, when necessary, should be efficient and fast. Large deposits can be removed mechanically, but smaller deposits are generally removed by gasification with steam or oxygen: C+HzO-CO+H,
(16)
c + 02
(23)
-
co2
Depending on the nature of the surface on which coke is deposited, these reactions may be catalysed or not. Reaction (23) is highly exothermic and considerable care must be taken when regenerating with oxygen. Catalysts may be permanently deactivated by sintering (loss of surface area) and this is known to start to be significant at temperatures between 33% and 50% of the melting points of the solids [41]. Inspection of Table 3 shows that maximal temperatures can be faily low, and careful control of burn-off is necessary. This is usually achieved by a slow step-wise increase in the temperature and the amount of oxygen fed to the bed. Thus, for example, a reforming catalyst may be regenerated by heating at 475 “C (oxygen slowly increased from 1% to 5% over about 24 h) and then at 510 “C for 2 h under 21% oxygen [27]. At all times, the temperature is carefully controlled so as not to exceed the critical values for platinum or alumina. Oxygen is the preferred gasifying agent in many cases since steam can accelerate sintering [42]. The steamcarbon reaction is easier to control, but metal oxides, in particular, sinter more readily in steam. This may even
present problems during gasification with oxygen, since partially hydrogenated residues will produce steam, and efficient product removal is desirable. In other cases, however, the effect of steam is less and the sensitivity of the system to over-temperature is greater, leading to preferred gasification with steam. Raney copper based catalysts are a case in point [43]. Recovery of heat liberated during coke removal is one of the few positive factors about regeneration, and in some cases it can be critical to the whole operation. Catalytic cracking is a case in point, where cracking is endothermic and uses the heat generated by coke removal [2]. In this case, it is important to recover all of the heat, and partial combustion 2c+o*-2co
(25)
is undesirable. As a result, it has become recent practice to add a trace of platinum to the silica-alumina [2], the metal having little effect on cracking but promoting complete oxidation: 2co + 02 -
2co*
(26)
The removal of coke from reactors also poses problems, but for different reasons. The deposits are usually thicker and, as a result, the catalytic action of the underlying metal is minimal. This results in the necessity of gasifying coke at higher temperatures, and steam cracker tube regeneration is usually carried out at about 600 “C rising to 800 “C. The initial gasifying agent is steam, an eventual mixture of steam and air (-50% of each) being introduced slowly. In these cases, sintering is less of a problem but carburization and metal dusting must be prevented. Overtemperatures increase carburization, a process that should be limited where possible. Metal dusting may occur when coke deposits within the alloy are gasified,
148 leading to physical separation of metal particles from the tube. There may be a case for gasifying only that carbon deposited on the alloy (and not the carbon dissolved in the alloy), but this is open to question since it would leave the tube in a carburized state. Studies are currently in progress to determine optimal decoking/recoking procedures.
7. Conclusions The formation of coke on catalysts and reactors is an unwanted side process that originates from several sources. Reactions in the gas and liquid phase and on the surfaces of solids may produce coke, and this can deactivate catalysts, increase heat transfer, block reactors and tubes, and cause high temperature corrosion. Minimization of coking can be achieved by careful control of individual systems. The thermodynamics of the desired and coking reactions dictate optimal operating conditions. Gas phase coking can be prevented by increasing dilution and surface:volume ratios. Catalytic coke control largely involves the catalyst designer/ manufacturer, but the chemical engineer may choose suitable catalysts to minimize the effect. Carburization may be reduced by careful choice of materials and by coating the surfaces with an inert material. Coking cannot normally be eliminated. Coke removal may be achieved mechanically or by gasification with oxygen or steam. Care is necessary to avoid high temperatures during the latter process.
References 1 0. Weisser and S. Landa, St&hide Catalysts: Their Properties and Applications, Pergamon, Oxford, U.K., and Viewig, Brunswick, F.R.G., 1973. 2 B. C. Gates, J. R. Katzer and G. C. A. Schuit, Chemistry of Catalytic Processes, McGraw-Hill, New York, 1979. 3 B. H. Thompson and C. T. Brooks, 165th Am. Chem. Sot. Natl. Meet., Dallas, TX, 1973. 4 P. S. Virk, L. E. Chambers and H. N. Woebcke, Adv. Chem. Ser., I31 (1974) 237. 5 S. M. Davis, F. Zaera and G. A. Somojai, J. Catal., 77 (1982) 439. 6 D. L. Trimm and C. J. Turner, J. them. Tech. Biotechnol., 31 (1981) 285. 7 S. C. Graham, J. B. Homer and J. L. J. Rosenfeld, Froc. R. Sot., Ser. A, 344 (1975) 259. 8 S. C. Graham,fioc. R. Sot., Ser. A, 377 (1981) 119. 9 D. L. Trimm, Cstal. Rev. Set. Eng., 16 (1977) 155. 10 J. G. McCarty and H. Wise,J. Catal., 57 (1979) 406.
11 J. R. Rostrup-Nielsen and D. L. Trimm, J. CataL. 48 (1977) 1.55. 12 W. G. Appleby, J. W. Gibson and G. M. Good, Ind. Eng. Chem. Process Des. Dev., I (1962) 102. 13 J. B. Butt and R. M. Billamora, Chem. React. Eng. Rev., Am. Chem. Sot. (Houston), (1978) 288. 14 R. Pearce and W. R. Patterson, Catalysis and Chemical Processes, Blackie, London, U.K. 1981. 15 L. F. Albright, B. L. Crynes and W. H. Corcoran, Pyrolysis. Theory and Industrial Practice, Academic Press, New York, 1983. in S. A. Jansson and Z. A. Foroulis (eds.), 16 1. Koszman, High Temperature Gas-metal Reactions in Mixed Environments, Met. Sot., A.I.M.E., New York, 1973, p. 155. 17 C. A. Bernard0 and D. L. Trimm, Carbon, 14 (1976) 225. Catal, 62 (1980) 35. 18 B. J. Cooper and D. L. Trimm,J. 19 D. L. Trimm and C. J. Turner, J. Chem. Tech. Biotechnoi., 31 (1981) 195. 20 N. M. Laurendeau, Frog. Energy Combust. Sci.. 4 (1978) 221. 21 N. N. Semenov, Some Problems of Chemical Kinetics and Reactivity, Pergamon, Oxford, U.K., 1958. 22 G. C. Bond, Catalysts by Metals, Academic Press, New York, 1962. Catal., 5 (1983) 263. 23 D. L. Trimm,Appl. 24 J. R. Rostrup-Nielsen, J. (7ataZ.. 25 (1984) 31. 25 J. J. McCarroll, Surf. Sci, 53 (1975) 297. 26 D. Briggs, J. Dewing, A. G. Burden, R. Moyes and S. P. B. Wills, J. Catal., 65 (1980) 31. 27 D. A. Dowden, Catalysis, Spec. Period. Rep. Chem. See., 2 (1978) 1. 28 R. T. Rewick, P. R. Wentrcek and H. Wise, Fuel, 53 (1974) 274. 29 C. A. Bernard0 and D. L. Trimm, Carbon, 17 (1979) 115. 30 J. R. Rostrup-Nielsen, Steam Reforming, Teknisk Forlag, Copenhagen, Denmark, 1975. 31 J. R. H. Ross, Surface and defect properties of solids, Spec. Rep. R. Sot. Chem., 4 (1975) 34. 32 R. T. K. Baker and R. D. Sherwood, J. CataL, 61 (1980) 378. 33 S. J. Tauster and S. C. Fung, J. Catal.. 55 (1978) 29. 34 D. J. Young, G. M. Smith and D. L. Trimm, Proc. 35th Aust. Inst. Metals Conf., Sydney, Australia, 1982, p. 86. 35 A. Schnaas and H. J. Grabke, Oxid. Met., I2 (1978) 387. 36 K. Ledjeff, A. Rahmel and M. Schorr, Werhst. Korros., 31 (1980) 38. 37 A. HoJmen, 0. Lindvag and D. L. Trimm, J. Chem. Tech. Biotechnol., 31 (1981) 311. 38 W. Karcher and P. Glaude, Carbon, 9 (1971) 617. 39 R. H. Long, M. C. Sze and H. Under, Met. Prog. (1975) 52. 40 I. Kozman, U.S. Patents 3 546 317 and 3531394,197O. 41 D. L. T&m, Design of Industrial Catalysts, Elsevier, Amsterdam, 1980. 42 D. L. Trimm, D. J. Young and P. Udaja, m B. Delmon and G. F. Froment (eds.), Studies in Surface Science and Catalysis, 1980, p. 331. 43 W. L. Marsden, M. S. Wainwright and J. B. Friedrich, Znd. Eng. Chem. Prod. Res. Dev., 19 (1980) 551.