20.2 Corrosion Protection of Turbine Blades

The development of turbine blade materials to meet the challenge of increasing engine temperatures must satisfy the requirements listed in Table 20.3.

These requirements severely limit our choice of very resistant materials. For example, ceramics, with their high softening temperatures and low densities, are ruled out for aeroengines because they are far too brittle (they are under evaluation for use in land‐based turbines, where the risks and consequences of sudden failure are less severe). Cermets offer no great advantage because their metallic matrices soften at a too much low temperature. The materials that best fill present needs are the nickel‐based superalloys.

The alloy used for turbine blades in the high pressure stage of an aircraft turbofan engine is a classic example of a material designed to be resistant to dislocation (power‐law) creep at high stresses and temperatures. At takeoff, the blade is subjected to stresses approaching 250 MN m⁻², and the design specification requires that this stress be supported for 30 hours at 850 °C without more than 0.1% irreversible creep strain. In order to meet these stringent requirements, an alloy based on nickel has evolved with the rather mind‐boggling specification given in Table 20.4. The point of all these complicated additions of foreign atoms to the nickel is straightforward. It is (i) to have as many atoms in solid solution as possible (the cobalt, the tungsten, and the chromium); (ii) to form stable, hard precipitates of compounds like Ni₃Al, Ni₃Ti, MoC, and TaC to obstruct the dislocations; and (iii) to form a protective surface oxide film of Cr₂O₃ to protect the blade itself from attack by oxygen. These superalloys resist creep so well that they can be used at 850 °C, and they are so hard that cannot be machined easily by normal methods and must be precision‐cast (for example, by investment casting that produces a fine‐grained material that may undergo a fair amount of diffusion creep and that may fail rather soon by cavity formation) to their final shape. Due to this, the blades are expensive, the total cost of a rotor of 102 blades being €200 000. To stave off failure, the alloys are directionally solidified (DS) to give long grains with grain boundaries parallel to the applied stress. DS alloys are standard in high performance engines and are now also in use in civil aircrafts. The improved creep properties of the DS alloy will allow the engine to run at a flame temperature approximately 50 °C higher than before, for a doubling in production cost. Figure 20.1 shows how this evolutionary process has resulted in a continual improvement of creep properties of nickel alloys over time and shows how the amounts of the major foreign elements have been juggled to obtain these improvements – keeping a watchful eye on the remaining necessary properties. The figure also shows how improvements in alloy manufacture – in this case the use of directional solidification – have helped to increase the operating temperature. Nevertheless, it is clear from the graph that improvements in nickel alloys are now nearing the point of diminishing returns.

Internal cooling of the blade enabled the inlet temperature of the turbine to be increased immediately by 100 °C with no change in alloy composition. Later, film cooling constituted a continuous improvement in turbine blades (Figure 20.2). But by further ducting cold air through the blades will begin to reduce the thermal efficiency, and thus a unique characteristic of some alloys, called eutectics, was explored. These eutectics form spontaneously on aligned reinforced structure, which is usually a compound with high melting temperature, whose fibers improve the creep properties of the resulting composite alloy.

Figure 20.3 shows how eutectics are made, and Table 20.5 lists some typical high temperature composites under study for turbine blade applications.

As shown in Figure 20.2, if directionally solidified eutectics (DSEs) prove successful, they will allow the metal temperature to be increased by ≈100 °C above conventional DS nickel alloys and the inlet temperature by ≈200 °C (because of a temperature scaling effect caused by the blade cooling). Further improvements in alloy design are underway in which existing nickel alloys and DSEs are being blended to give a fiber‐reinforced structure with precipitates in the matrix.

The ceramics best suited for structural use at high temperatures (≥1000 °C) are listed in Table 20.6 and compared with nickel‐based superalloys. The comparison shows that all the ceramics have attractively low densities, high moduli, and high melting points (and thus excellent creep strength at 1000 °C); and many are completely resistant to oxidation – they are oxides already. But several have poor thermal conductivity (leading to high thermal stresses), and all have very low toughness.

To overcome their major problems, ceramic matrix components are under development: they combine strong, exceptionally perfect, fibers (such as silicon carbide or alumina, grown by special techniques) in a matrix of ceramics (silicon carbide or alumina again), made by conventional means to give a material that you might think of as “high temperature wood.”

Any major materials development program, such as that on the DSEs, can only be undertaken if a successful outcome would be cost effective. The cost of the materials themselves must also be considered. For example, the use of greater quantities of exotic materials, such as hafnium, will drive the cost of blades up. But if new alloys offer improved life or inlet temperature, there is a strong incentive to pursue them.

As we saw above, the alloys used for turbine blades contain large amounts of chromium, dissolved in solid solution in the nickel matrix. Now, if we look at Table 20.1, we see that the formation of Cr₂O₃ releases much more energy (701 kJ mol⁻¹ of O₂) than NiO (439 kJ mol⁻¹ of O₂). This means that Cr₂O₃ will form in preference to NiO on the surface of the alloy. Obviously, the more Cr in the alloy, the greater the preference for Cr₂O₃. At the 20% level, enough Cr₂O₃ forms on the surface of the turbine blade to make the material act a bit as if it were chromium. Suppose for one moment that our material is chromium. Table 20.2 shows that Cr would lose 0.1 mm in 1600 hours at 0.7 T_M. Of course, we have forgotten one thing, 0.7 T_M for Cr is 1504 K (1231 °C), whereas, as already mentioned, for Ni it is 1208 K (935 °C). We should, therefore, consider how Cr₂O₃ would act as a barrier to oxidation at 1208 K rather than at 1504 K (Figure 20.4). The oxidation of Cr follows parabolic kinetics with an activation energy of 330 kJ mol⁻¹. Then the ratio of the times required to remove 0.1 mm (from equation , where A_p and Q_p are constants) is

20.1

Thus, the time at 1208 K is t₂ = 0.65 × 1.6 × 10⁶ hours = 1.04 × 10⁶ hours. Note that in Eq. 20.1, t₁ and t₂ are the oxidation times with t₂ > t₁.

Now, as we have said, there is only at most 20% Cr in the alloy, and the alloy behaves only partly as if it were protected by Cr₂O₃. In fact, experimentally, we find that 20% Cr increases the time for a given metal loss by only about 10 times, i.e. the time taken to lose 0.1 mm at blade working temperature becomes 600 × 10 hours = 6000 hours rather than 10⁶ hours.

Why this large difference? Well, whenever you consider an alloy rather than a pure material, the oxide layer – whatever its nature (NiO, Cr₂O₃, etc.) – has foreign elements contained in it, too. Some of these will greatly increase either the diffusion coefficients in, or electrical conductivity of, the layer and make the rate of oxidation through the layer much more than it would be in the absence of foreign elements contamination.

This 0.1 mm loss in 6000 hours from a 20% Cr alloy at 950 °C, though better than pure nickel, is still not good enough. The obvious way out of this problem is to coat the blades with a protective layer (Figure 20.5). This is usually carried out by spraying molten droplets of aluminum onto the blade surface to form a layer, some microns thick. The blade is then heated in a furnace to allow the Al to diffuse into the surface of the Ni. During this process, some of the Al forms compounds such as AlNi with the nickel – which are themselves good barriers to oxidation of the Ni – while the rest of the Al becomes oxidized up to give Al₂O₃.

Because oxides are usually quite brittle at the temperatures encountered on a turbine blade surface, they can crack, especially when the temperature of the blade changes and differential thermal contraction and expansion stresses are set up between alloy and oxide. These can act as ideal nucleation centers for thermal fatigue cracks, and because oxide layers in nickel alloys are well stuck to the underlying alloy (they would be useless if they were not), the crack can spread into the alloy itself.

The properties of the oxide film are thus very important in affecting the fatigue properties of the whole component. The pure refractory metals Nb, Ta, Mo, and W have high melting temperatures (2740, 3250, 2880, and 3680 K, respectively) and should therefore have very good creep properties. But they oxidize very rapidly indeed (see Table 20.2) and are utterly useless without coatings. The problem with coated refractory metals is that if a break occurs in the coating (e.g. by thermal fatigue, erosion by dust particles, etc.), catastrophic oxidation of the underlying metal will take place, leading to rapid failure. The “unsafeness” of this situation is a major problem that has to be solved before we can use these on‐other‐counts potentially excellent materials.

The ceramics SiC and Si₃N₄ do not share this problem. They oxidize readily (Table 20.1); but, in doing so, a surface film of SiO₂ forms that gives adequate protection up to 1300 °C. And because the film forms by oxidation of the material itself, it is self‐healing.

20.3 Oxidation of Silicides for VLSI Applications

Electronic materials are those used in electrical industries, electronics and microelectronics, and the substances for the building up of integrated circuits (ICs), circuit boards, packaging materials, communication cables, optical fibers, displays, and various controlling and monitoring devices. Discovery, development, and application of new materials are the robust power for the development of human society.

Processing of electronic materials mainly includes crystal growth, epitaxy, lithography, deposition, annealing, doping (diffusion or ion implantation), etching (dry or wet), and metallization. In ICs, the fundamental electronic switching device is the (complementary) metal oxide semiconductor field effect transistor (MOSFET), which has at least three terminals – the gate, source, and drain. The gate electrode is separated electronically from the source and drain by a thin dielectric film, which is commonly SiO₂. Formation of this film could be accomplished by using sputtering, evaporation, chemical vapor deposition, or atomic layer deposition. Thermal oxidation is also used to grow the gate dielectric films due to its ease in processing control, but it results in many problems in association with material degradation at high temperature; these problems lead to the formation of oxides, carbides, sulfides, nitrides, etc., but when these oxidation processes are well controlled, they have been finding many applications in the modern electronics industry for surface cleaning, film growth, preparation of new materials, and building up of novel structures for devices. To date, Si is still the backbone of the modern semiconductor industry due to its (i) availability in a wide variety of sizes and shapes, (ii) mature material preparation and property control, (iii) native oxide films on its surface, and (iv) compatibility to planar IC technology.

Silicides of transition metals are used in two distinct ways in very large scale integrated (VLSI) circuits: (i) in the contact pads of individual devices on a chip and (ii) in the interconnection lines. For the latter applications, two particular properties of the silicides make them desirable: their metallic electrical conductivity and the possibility to produce a protective coating of SiO₂ under heat treatment in an oxidizing ambient. Of those two attributes, the second one is particularly attractive because it offers inherent advantages in multilevel metallization schemes. To use this feature of silicides successfully, an understanding of the oxidation process is needed. It is convenient to subdivide the subject into two parts, depending on whether the silicide rests on a substrate of excess silicon or whether the substrate is inert, that is, the substrate is in a fully oxidized state, such as SiO₂. Other reasons for the use of silicides in VLSI applications are their very low resistance; good process compatibility with Si, e.g. ability to withstand high temperatures; oxidizing ambients; various chemical cleans used during processing; little or no electronization; easy to dry etch; and good contacts to other materials. But these benefits rise many problems in integrating silicides in an IC, and under this point of view, the actual preferred silicides are WSi₂, TiSi₂, NiSi, and CoSi₂. Table 20.7 indicates preferred silicides for VLSI application. The main techniques used for the formation of these silicides are (i) metal deposition on Si and formation by thermal heating, laser irradiation, or ion beam mixing (these are sensitive to interface cleanliness and heavy doping, selective silicidation on Si possible, and widely used for silicides of Pt, Pd, Co, and Ti), (ii) co‐evaporation (E‐gun) of metal and Si (poor process control, poor step coverage, and good tool for research but not used in manufacturing), and (iii) sputtering from a composite target (possibility of high level of contaminants like C, O, Na, and Ar; poor step coverage; used for MoSi₂ and WSi₂).

20.3.1 Oxidation of Silicides on Silicon

Typically, silicides are not used alone as interconnections, but with an underlayer of polycrystalline Si. Upon thermal treatment in an oxidizing ambient, most silicides form a SiO₂ overcoat while the silicide layer below is morphologically preserved. The rate of growth of the SiO₂ layer is different for the different silicides, and these rates also differ from that of pure Si. They can be represented by the linear–parabolic growth law

20.2

where x is the oxide thickness, t the oxidation time, B the parabolic rate constant, B/A the linear rate constant, and τ a fitting parameter that takes into account the initial transient oxide growth. Table 20.8 lists the growth kinetics of silicides. There are some discrepancies between different studies of the same silicides, but the overall differences between various silicides cannot be explained by details of the silicide preparation method. What then is the origin of these differences? An analysis of the results leads to the following observations:

1. The parabolic rate constant is the same for constant oxidizing conditions, regardless of the substrate (single crystalline Si, polycrystalline Si) or the silicide. The rate constant is equal to that observed for pure Si under similar oxidizing conditions (wet or dry). Variations in the partial pressure of water vapor can explain the remaining differences for wet oxidation. The conclusion is that the diffusion process through the oxide is the same for all combinations of Si substrates and silicides. This also implies that the physical properties of the oxide are similar. The dissimilarities in the oxidation rates listed in Table 20.8 thus originate in the linear terms of the growth law.
   
2. The linear term in the growth law can be related with three processes: the mass transport by diffusion across the silicide of fixed thickness and the reactions of the two interfaces of the silicide layer, i.e. the formation of SiO₂ at the silicide/SiO₂ interface, and the transfer of Si from the Si substrate into the silicide layer at the Si/silicide interface. This latter process is similar in nature to the reaction that takes place during the formation of a silicide by thermal annealing of a metallic film with a Si substrate. In those reactions, at low temperatures at which they can be analyzed in detail (∼300–600 °C), the transport across the growing silicide layer is the rate‐limiting step in all cases where the growth law has been firmly established (with the exception of CrSi₂, where the growth is clearly linear in time). This says that the transfer of Si from the substrate to the silicide is not the rate‐limiting step. At the low temperatures at which these facts are established (for the formation of a silicide by thermal annealing), the rate of the Si consumption is of the same order as that observed during the oxidation process listed in Table 20.8. But at those rates, the reaction at the Si/silicide interface is not rate limiting already at low temperatures; thus, it must be negligible also at the high temperature during oxidation. The cause for the dissimilar linear rate constants in the oxidation process is thus connected with differences in the oxidation process of the various silicides at the silicide/SiO₂ interface, with differences in the transport across the silicide or with both.
   
3. The observed differences in the oxide thickness caused by dissimilarities in the linear (initial) part of the oxidation can be correlated with the process of atomic transport through the silicide layer. Table 20.9 compares the dominant moving species for silicide formation by thermal reaction with a metal film and for oxidation. They are the same, with the exception of the special cases of Cr and Pd. On the assumption that only one species dominates the transport, there are two distinct limiting cases that can be distinguished, according to whether Si or metal moves.
   

   When the metal moves, it must be supplied from the dissociation of the silicide at the silicide/SiO₂ interface as a result of the oxidation process. At the elevated temperatures of oxidation, the metal transport process is quite fast, as it is known from silicide formation reactions with metal films, whose rates are similar at lower temperatures. If the dissociation process at the silicide/SiO₂ interface is very effective, the oxidation process will be controlled only by the oxygen arrival at the interface, and the linear term in the growth equation should be negligible. Experiments indeed confirm this scenario for CoSi₂ and PdSi. When the dissociation process is so effective that it produces an excess of weakly bonded metal and Si at the interface, the possibility of an additional transport of Si down to the substrate, in parallel to the transport of metal, also arises. In this case, both atomic species participate in the mass transport. This situation has been identified for CrSi₂, NiSi₂, and PdSi. Note that the cases of CoSi₂ and PtSi differ from these three cases only quantitatively in that the Si transport may be too small to be measured for CoSi₂ and PdSi. Qualitatively, these five silicides are all similar. One would expect that these conditions of ready dissociation and rapid mass transport through the silicides favor the highest possible oxidizing rates. Indeed, the silicides that dissociate during oxidation (CoSi₂, NiSi₂, PdSi, PtSi) have the highest oxidation rates. Note that the lowest eutectic temperature in the Pt–Si system is 830 °C; oxidation studies performed at 750 °C yield a combined linear and quadratic growth law. From the point of view of low temperature processing, the near‐noble metal silicides are thus preferred candidates for IC interconnections. These results are summarized in Table 20.10. It is seen that there are two additional transport cases that are conceivable when both species move and the silicide layer is preserved during oxidation. These cases have not been seen experimentally.

   
   

   The second limiting case is represented by TiSi₂. Silicon is the main diffusion species there. The transport of Si across TiSi₂ is one of a few cases where the diffusion mechanisms have been identified for the silicide formation by reaction with a Ti film using radioactive tracers. The Si diffuses from the substrate by a process that involves the atoms in the silicide lattice (e.g. by Si vacancies). In this case, dissociation of the silicide needs to occur. In the initial phase of the oxidation, where the supply of oxygen is high because the oxide layer is thin, the reaction at the silicide/SiO₂ interface reduces the overall rate of oxidation, and the linear rate constant B/A begins to be important as well. As a result, the growth is slowed down initially as compared with the previous limiting case.

   
   

   There are uncertainties in the reported values of the linear rate constant for different silicides, because the measurements are difficult to be done accurately. It is nonetheless clear that for both wet and dry oxidation, the linear rate constant is higher for silicides than it is for silicon. The observation can be linked to the metallic properties of the silicides. An electron transfer process in one or several of the steps of the oxidation reaction at the interface can readily alter and accelerate the reaction. That electron transfer process is more probable when the surface is metallic than when semiconducting. Support for this idea is provided by the wet oxidation of Ir₃Si₅. This silicide is known to be a semiconductor. Its linear rate constant is indeed larger than that of silicon but smaller than that of the metallic silicides. The actual processes that determine the value of the linear rate constant for oxidation of different silicides on silicon remain to be established. This subject is one of the main open issues of silicide oxidation that awaits clarification.

   
   
4. Thermodynamically, the formation of pure SiO₂ is the expected outcome for all but a few silicides. The available thermal data are uncertain and incomplete. From estimates of the standard heats of reaction, only ZrSi₂ and HfSi₂ are not expected to form SiO₂, but metal oxide instead, and TiSi₂, NbSi₂, and TaSi₂ are marginal cases. HfSi₂ and NbSi₂ have indeed been reported to disintegrate and to form a mixture of metal and Si oxides. TiSi₂ forms SiO₂ also but seems to do so only above 900 °C. Recent ternary phase diagrams for Ti, Ta, Mo, and W also predict that with excess Si, SiO₂ is in equilibrium with the disilicides of these metals. The general conclusion is that the thermodynamical argument correctly predicts the general oxidation behavior and identifies the marginal cases. The appearance of SiO₂ is thus not the result of kinetic limitations at the silicide/SiO₂ interface, but indicates instead that the reaction is there controlled by constraints of thermal equilibrium.

a Units of B* are in centimeters and seconds with appropriate powers.

Source: Adapted from Bartur and Nicolet (1984).

Silicide	Dominant diffusion species during the following processes
Silicide formation	Silicide oxidation
TiSi2	Si	Si
CoSi2	—	Co
NiSi2	Ni	Ni
Pd2Si	Si, Si and Pd	Si
PdSi	Pd and Si	Pd
PtSi	Pt	Pt
CrSi2	Si	Cr

Source: Adapted from Bartur (1983).

Moving species during oxidation	Schematic description		Experimental results
		Si supply	TiSi2 Pd2Si
		Dissociation	CoSi2 PdSi
		“Excess” dissociation	CrSi2 NiSi2 PtSi
		Dissociation and Si supply
		Si supply causes metal release and diffusion
	These processes do not preserve the silicide

Source: Adapted from Bartur (1983).

These considerations suggest that silicides in which the metal is the dominant moving species offer an advantage over the others when low temperature oxidation is desired, but other considerations, such as the diffusion of metal atoms from the silicide into the Si substrate or the electrical resistivity of the silicides, are important in practical applications as well.

20.4 Naphthenic Acid Corrosion in Petrochemical Plants

New technologies and processes that are being applied in the petrochemical industry are requiring higher operating temperatures and pressures of the process units. Therefore, catastrophic consequences such as firing and/or explosion might be caused by the failures initiated by severe corrosion and oxidation attacks of the materials. The major high temperature corrosion attacks in the petrochemical industry include high temperature oxidation, high temperature sulfidation, high temperature naphthenic acid corrosion, and high temperature carburization. Extensive studies and tests on these problems at a laboratory scale have been reported, but the acquired knowledge is not sufficient to guide field corrosion protection practice. Thus the rational selection of materials for the construction of processing units based on the analyses of the nature of the materials and their actual operating conditions is the key to controlling high temperature corrosion.

Oxidation, sulfidation, and carburization are three types of phenomena that are treated in Chapters 7–9 of this book. Here, we will only consider two case studies of naphthenic acid corrosion.

Naphthenic acid in crude oils is a rather complex mixture, with a saturated ring structure and a molecular weight ranging from 180 to 700. Its content is usually expressed by the total acid number (TAN; as milligrams of KOH required to neutralize the acid in 1 g of oil).

During the refining processes, naphthenic acid is heated, distilled, vaporized, and condensed. Then it can dissolve into the distillate oil, resulting in corrosion problems. Particularly at temperatures higher than 240 °C, serious corrosion reactions take place:

20.3

20.4

Iron naphthenate, the product of naphthenic acid corrosion, is oleo‐soluble; therefore, the corroded metal surfaces are normally clean and free of scales. In the regions with high temperature and high flow velocity, naphthenic acid corrosion will result in the formation of shape‐edged streamline grooves along the flow direction. In the areas with low flow velocity, shape‐edged pits will be formed. Since Fe(COO)₂ can dissolve into oils and then be removed easily by the flowing media, it is difficult to form protective scales on the metal surface. Even FeS formed on the surfaces can react with naphthenic acid: fresh metallic surfaces will be exposed to the corrosive media, and corrosion can proceed continuously.

TANs of the crude oil and the distillate oil are the key factor to evaluate high temperature naphthenic acid corrosion. It is generally accepted that when the TAN is higher than 0.5 mg KOH g⁻¹, high temperature naphthenic acid corrosion will occur. Since naphthenic acid exists together with other oils that have similar boiling points, only the actual acid number of the distillate oils can determine the corrosivity and the corrosion rate. Under atmospheric distillation conditions, the highest acid number of the distillate oil will be shown in the temperature range of 371–426 °C and at 260 °C in vacuum distillation towers.

Two case studies in high temperature naphthenic acid corrosion that deserve description are those with high TAN and low sulfur and those with high TAN and high sulfur.

20.5 Oxidation of Ceramic Matrix Composites

The class of monolithic and composite ceramic materials termed engineering, technical, or advanced ceramics exhibits an attractive range of properties relative to metallic counterparts commonly used in a number of high temperature applications. In particular, retention of mechanical properties to temperatures in excess of 1200 °C presents the opportunity to plant designers to increase operating temperatures, improve process efficiencies, and reduce the levels of emitted pollutants. Lower densities of ceramic materials than competing metallic counterparts permit the achievement of higher specific strength and thus increased pay loads and reduced energy consumption. Advanced ceramics are thus attractive candidate structural materials for use in energy production and conversion and in engines. Continued progress to actual adoption in commercial installations has been hindered mainly by poor mechanical reliability and uncompetitive production costs. Advances in manufacturing technology are yielding improvements in mechanical performance and gradually reducing the cost of production. While the financial systems for determining cost are well developed and understood, the judgment of component reliability is dependent on the availability of a solid base of valid materials test data.

Independently of these issues, it is a fact that structural ceramics are finding more and more applications in high temperature systems, as shown in Table 20.11, and therefore oxidation and corrosion at high temperatures becomes an important field of study. Ceramics can be classified according to the type of protective oxide they form. These include silica formers (SiC, Si₃N₄), alumina‐forming non‐oxides (AlN, Al₄C₃), and boria‐forming ceramics (BN). Next, transition metal carbides and nitrides that form the corresponding transition metal oxides, as well as the ceramics that form the mixed oxide scales, are also deserving attention. Finally, the oxidation of ceramic matrix composites (CMCs), which involves several parallel processes, is an important current area of research. These CMCs are the most promising ceramics today for high temperature structural applications. They include whiskers, particulates, and continuous fillers in a ceramic matrix to impart improved fracture toughness. Most of the current research on CMCs is focused on SiC inert fiber coatings in matrices of oxides, Si₃N₄, or SiC. The most adequate coatings are graphite carbon and boron nitride. Thus, the entire composite has a slowly oxidizing phase (fibers and matrix) and an easily oxidizable phase (fiber coating). Generally, at high temperatures, SiO₂ rapidly forms on the outside of the matrix, and attack of the fiber coating is minimized. Even a matrix crack may quickly seal. However, at intermediate (500–1000 °C) temperatures, silica forms very slowly, and oxidative attack of the fiber coating becomes a critical issue. The most thorough treatment of this problem is from Filipuzzini (1994) for carbon‐coated SiC fibers in a SiC matrix. The initial treatment consisted of a matrix with unidirectional fibers with one end exposed. The process consisted on a two‐step oxidation of carbon and concurrent oxidation of the fiber and surrounding matrix. As carbon oxidizes, an annular region around the fiber is left. Filipuzzini first measured oxidation properties of each of the constituent parts of the composite as well as the composite as a whole. Standard thermo‐gravimetric analysis (TGA) methods and resistivity measurements gave an indication of the amount of carbon consumed. These data were used in an overall description of the composite oxidation. By introducing analytical models with appropriate boundary conditions, it was possible to generate TGA oxidation curves, indicating that thinner carbon coatings give better performance. This analysis has been extended to other situations. For example, the examination of the oxidation of SiC matrices with carbon fibers indicated that the low temperature oxidation is kinetically controlled by reaction of oxygen with the fibers and high temperature oxidation is diffusion controlled by diffusion of oxygen to the fibers. Finite difference models of this process were developed.

The other commonly used fiber coating is hexagonal BN. There are several important phenomena in the oxidation of a silicon‐based composite with BN fiber coatings. As the silicon‐based material and the BN oxidize, both B₂O₃ and the SiO₂ form. These react to form low‐melting borosilicate melts. Related to this, boron enhances the oxidation of silica‐forming materials. Also, water vapor readily reacts with B₂O₃ oxidation product and forms volatile H_xB_yO_z (g) species, which leads to effective volatilization of BN. However, unlike carbon volatilization, distinct interfaces are not formed with the BN. Rather there is a “plug” of borosilicate formed and a gradual transition from oxide to nitride. Jacobson, in 1999, described a model that explains the formation of this borosilicate plug, based on gas‐phase diffusion of the H_xB_yO_z (g) species outward and boron‐enhanced oxidation of the SiC. Exposures of the SiC fiber to BN fiber coatings in a SiC matrix were conducted at 700 and 800 °C in 1 and 10% H₂O/O₂ atmospheres with one end of the test sample ground off to expose the fiber ends and coatings. Then recession distances were measured, but exact distances were difficult to determine as interfaces were not always sharp. Therefore, an adjustable parameter of boron‐influenced SiC oxidation rates had to be introduced. Using substantially enhanced oxidation rates, reasonable agreement between the model and experimentally measured recession distances was obtained. Nowadays, the oxidation of CMCs continues to be a major issue because these ceramics are finding novel applications and are being exposed to more complex corrosive environments.

20.6 Shell Corrosion of Rotary Cement Kilns

The subject of reinforcement corrosion is widely described in the open literature, but descriptions of corrosion that occurred in dry‐process, rotary, cement‐clinker kilns and other structures producing cement seldom appear in literature. Here we describe observed corrosion of kiln shells reported by Potgieter, in 2004, which occurred at two rotary cement kilns, in South Africa.

A rotary cement kiln can be seen as a complex chemical reactor with various temperature zones and differing container wall constituents in the different temperature zones. Further on, the number of preheaters, raw material composition, and operating conditions can all influence the corrosion occurring in a rotary cement kiln. The chemical composition of the raw material used and type of refractory used are the main variables contributing to corrosion. Chlorine, sulfur, and alkalis are usually present in the kiln, leading to the production of low‐melting eutectics that, in combination with the oxidizing atmosphere that exists in the kiln, are aggravating factors that can further salt infiltration into the refractory and ultimately hasten corrosion that occurs on the kiln shell.

In practice, the observed corrosion processes are similar to those occurring in steam boiler walls that are covered with deposits; thus the kiln acts as a wall deposit. The iron from the steel shell, in contact with the oxidizing atmosphere, originates several iron oxides (Fe₃O₄ in particular) with hematite (Fe₂O₃) at the scale–brick interface. Sulfur from the fuel can react in the oxidizing atmosphere of the kiln to produce SO₂ and SO₃ acid gases. In the oxygen‐restricted environment between the refractory lining and the kiln shell, it is conceivable that SO₂ or SO₃ can act as the oxygen donors and react with the iron giving further magnetite (Fe₃O₄) and iron sulfides (troilite, FeS; pyrite, FeS₂).

The corrosion scales examined in this case study were removed from two shells of two kilns during maintenance when the dolomite bricks in the transition zones were replaced. The scales were brittle and porous with a layered brownish appearance and a few salt crystals. Kiln 1 was a long (140 m) dry‐process rotary furnace with only one preheater at the feed end; kiln 2 was a short (85 m) kiln furnace with five preheaters fixed at the feed end of the kiln. In both kilns, corrosion occurred in the transition zone at about 30 m from the coated clinkering zone. Chemical analyses of the corrosion products of the kiln shell showed high concentrations of both chloride and potassium, suggesting a chloride type of attack; the corrosion products from kiln 2 shell showed high sulfate and calcium concentrations, suggesting a sulfur‐dominant type of corrosion attack. The positive identification of CaSO₄ observed by X‐ray diffraction (XRD) could be explained by an excess of sulfur oxides present in kiln 2 atmospheres and their penetration through the refractory lining; it was essentially sulfidation corrosion. The other corrosion products showed elevated KCl concentration. Whereas chlorine gas can react with the kiln 1 shell, the alkalis can only penetrate the refractory brick lining as part of a salt melt that in rotary cement kilns is conceivable to occur at about 350–400 °C. Therefore, the situation is typical of hot corrosion processes. The severe problem is made worst by the practice, at this plant, of recirculating kiln and dust from the electrostatic precipitators, a dust that is rich in alkalis originating from raw materials of high alkali content. Potgieter and his collaborators recommended (i) to use a denser brick (e.g. magnesia–zirconia and magnesia‐enriched dolomite), (ii) to avoid the recycling kiln dust to the raw meal, and (iii) to consider a chloride or alkali bypass.

20.7 Corrosion of Steels in a Linear α Olefin Plant

Linear α olefins (LAOs) are a range of industrially important α olefins, which are commonly manufactured by oligomerization of ethylene and by Fischer–Tropsch synthesis followed by purification. HCl formation during α‐SABLIN (LAOs) commercial plant operation is the main reason for the production of organic chloride as well as of alkylated toluene by Friedel–Crafts alkylation reaction. The HCl and organic chloride affect the product quality originating gross corrosion at the separation section downstream. Consequently, the LAO plant is subjected to many inspection shutdown, which, in turn, affect the sustainability of the plant operation.

Steel materials, namely, carbon steel (CSA516‐70) and ferritic (SS410) and austenitic (SSS304L) stainless steels, are currently used in the construction of the α‐SABLIN commercial plant, and, therefore, Amin et al., at Saudi Arabia and Egypt (2014), published a case study on this issue, comparing the high temperature corrosion behavior of these three alloys in an autoclave at 270 °C and 29 bar to select the proper alloys to be used as a construction material for the most vital parts of the SABLIN’s plant. Alloy samples and plant solution were prepared by traditional means and then subjected to immersion tests of different duration (3–30 days). Corrosion rates were calculated by thermogravimetry, and the corrosion products were analyzed by OM, SEM, and XRD. It was observed that at any given immersion time, the corrosion rate of the three tested samples decreased following the order CSA516 > SS410 > SS904L. It was also revealed that the average corrosion rates of CSA516 and SS410 follow the kinetics of zero‐order reaction, up to 14 days of immersion, deviation from zero‐order kinetics being quite clear at larger immersion (30 days of immersion). For SS304L, the kinetics of zero‐order reaction was observed over the entire immersion time, as the alloy’s surface is almost free from corrosion products, even after 30 days of immersion in the corrosion test solution. XRD patterns of the corrosion products of alloys CSA516, SS410, and SS304L displayed the presence of goethite (α‐FeOOH), FeCl₂, Fe₃O₄, and γ‐Fe₂O₃ in the case of CSA516; goethite (α‐FeOOH), FeCl₂, Fe₃O₄, γ‐Fe₂O₃, Fe, and Fe₂O₃/(Fe,Cr)₃ in the case of SS410; and goethite (α‐FeOOH), Fe₃O₄, Fe, Fe₂O₃/(Fe,Cr)₃O₄, and (Fe, Ni) in the case of SS304L. FeCl₂ was not detected by the XRD of SS304L, meaning that the corroded surface is free of it, that is, its corrosion resistance is higher. The optical etched image of CSA516 showed the presence of ferrite and pearlite multiphase structure. The presence of ferrite grains and pearlite grains (which are also mechanical mixtures of lamellar ferrite and lamellar iron carbide or cementite) creates galvanic cells, thus enhancing the corrosion process. The chromium in SS410 (>12%) and SS304L (>18%) combines with oxygen in the atmosphere to form a thin, invisible layer of chrome‐containing oxide passive film, as confirmed from the formation of (Fe,Cr)₃O₄ phase. The high carbon content of alloy SS410 (C = 0.15%), as compared with that of SS304L (C = 0.03%), gives a real chance for this alloy to form Fe and Cr carbides in its microstructure. These carbides also decrease the corrosion resistance of alloy SS410 as compared with SS304L. Furthermore, the low Cr and Ni contents of alloy SS410, in addition to its high carbon content, as compared with SS304L, are the most likely reasons why SS410 recorded corrosion rates higher than those of SS304L. SEM images showed that the corrosion products accumulate on the surfaces of the carbon steel and the ferritic stainless steel, the austenitic stainless steel being free from corrosion products. But shallow pits were observed on its surface, probably due to the presence of the aggressive chloride ions. In summary, it was found that SS304L recorded the highest corrosion resistance among the tested samples, taking into account its high Cr and Ni contents and the high stability of (Fe,Cr)₃O₄. The high C content and low Cr and Ni contents are the reasons why SS410 recorded corrosion rates higher than those of SS304L. The presence of ferrite grains in contact with pearlite grains in the microstructure of CSA516 creates galvanic cells that explain its low corrosion resistance. The roughness of the surfaces of the three tested samples, examined by OM/SEM, justified the previous findings.