MATERIAL- OCH KEMITEKNIK 1232
Smooth Surfaces: A review of current and planned smooth surface technologies for fouling resistance in boiler
Robert Corkery, Linda Bäfver, Kent Davidsson, Adam Feiler
Smooth Surfaces: A review of current and planned smooth surface technologies for fouling resistance in boiler.
Glatta ytor: En genomgång av glatta ytor som teknik för att stå emot bildning av beläggningar på värmeöverförande ytor.
Robert Corkery, Linda Bäfver, Kent Davidsson, Adam Feiler
M08-838
VÄRMEFORSK Service AB 101 53 STOCKHOLM · Tel 08-677 25 80 Februari 2012 ISSN 1653-1248
VÄRMEFORSK
Abstract
Here we have described the basics of boilers, fuels, combustion, flue gas composition and mechanisms of deposition. We have reviewed coating technologies for boiler tubes, including their materials compositions, nanostructures and performances. The surface forces in boilers, in particular those relevant to formation of unwanted deposits in boilers have also been reviewed, and some comparative calculations have been included to indicate the procedures needed for further study. Finally practical recommendations on the important considerations in minimizing deposition on boiler surfaces are made.
i VÄRMEFORSK
ii VÄRMEFORSK
Sammanfattning Det höga innehållet av alkali (kalium och natrium) och klor i avfall och biobränsle leder till drift- och underhållsproblem genom bildning av påslag och efterföljande korrosion på överhettare i förbränningsanläggningar. Detta projekt handlar om interaktioner på överhettartubers ytor vid förbränning av avfall och biobränsle. En genomlysning av tekniker som kan användas för att skapa glatta tubytor i pannor presenteras. Glatta ytor skall förhindra eller fördröja bildning av påslag.
Påslag består av askelement från bränslet. Vid överhettare är kaliumklorid särskilt kritiskt och vid avfallsförbränning även natriumklorid. Både kalium- och natriumklorid är korrosiva. Vid avfallsförbränning kan även ämnen som bly och zink vara betydelsefulla. Hög koncentration av svavel i bränsle och därpå följande rökgas är istället positivt eftersom alkalisulfater kan bildas och de är inte lika klibbiga och korrosiva som motsvarande klorider. Både gasformiga ämnen och partiklar kan bilda påslag. De grundläggande mekanismerna för bildning av påslag är: tröghetsimpaktion (partiklar > 1 µm), diffusion (gas eller partiklar < 1 µm), termofores (partiklar < 1 µm), kondensation (gas) och kemiska reaktioner (gas eller partiklar). Vid överhettare (rökgastemperatur 800-1000 °C) är kalium- och natriumklorid i gasfas. Närmast överhettarytorna blir dock temperaturen lägre och kalium- och natriumklorid övergår till fina partiklar eller kondenserar direkt på ytorna. Ett sådant påslag kan vara tunt och homogent, medan andra påslag kan bli 10-15 cm tjocka och vara heterogent sammansatta.
Man kan minska påslag genom att belägga en ståltub med en beläggning som är keramisk eller med keramisk kompositbeläggning, så att ytan blir glatt. Den stora utmaningen vid keramisk beläggning av ståltuber är skillnader i termisk utvidgning av keramiskt och metalliskt material vid uppvärmning. Det leder till sprickor eller termisk chock med avskalning och andra problem. Cykling av tubtemperaturer vid drift kan också påverka gränsytor mellan ståltub och beläggningsmaterial. Olika keramiska material kan skräddarsys för att minimera termisk utvidgning (CTE - Coefficient of Thermal Expansion) och medföljande problem, men än så länge är de inte tillräckligt inerta i förbränningsmiljöer.
Termisk chock är oundviklig, men konsekvenserna kan minimeras genom att sprida ut spänningar. Det kan göras genom att med hjälp av legeringslager binda keramiskt material till tubmaterialet. Man kan använda komposit-keramer vilka innehåller partikulärt material med modifierad CTE och mer nanostrukturerade system som har mycket fina fibrer för homogen spänningsöverföring, mineralfibernätverk för flexibilitet och styrka, och till slut få en viss nivå av porositet som klarar volymförändringar utan att krackelera. I beläggningsmaterial som står emot termisk chock väl har nano- och mikrostrukturer en avgörande betydelse.
När man gör keramiska beläggningsmaterial kompatibla med stål uppstår oundvikligen porositet, nano- och mikrostrukturer som gasformig alkaliklorid kan diffundera in i. Det kan orsaka problem om struktursystemen är sammankopplade, vilket sannolikt är fallet i dagens material. Det kan också uppstå problem med porer och ledning av alkali-ånga vid tillverkning av keramiska beläggningsmaterial med slamgjutning.
iii VÄRMEFORSK
Fördelningen av porer i en keramisk beläggning som står emot termisk chock innebär sannolikt nanoporös struktur. Kelvin-ekvationen säger att KCl kondenserar in i dessa porer, väl under mättnadstryck för KCl i gasbulken, så vätskeformig KCl stabiliseras även vid högre temperaturer. Vidare så stabiliseras dessa vätskor av små porer. På så sätt stabiliseras vätskor vid mycket lägre temperaturer i den keramiska beläggningen på panntuben. Sammanfattningsvis så kondenserar KCl i porer med hjälp av kapillär kondensation, trots att KCl är i gasfas vid den aktuella rökgastemperaturen och de små porerna kommer också hjälpa till att transportera KCl mellan keramisk beläggning och ståltub.
Ytkrafter gås igenom utgående från beskrivna utmaningar för att belägga ståltuber med keramiskt material. Slutsatsen är att bästa metoden för att minimera påslag och korrosion med hjälp av keramisk beläggning, och upprätthålla livslängden för en keramisk beläggning är att använda keramisk nanokomposit som är resistent mot termisk chock med ”tie layer” till ståltub. Mer arbete behövs för att förstå den exakta ytkemin för alkaliklorid och interaktioner i sprickor och mikrostrukturer med hänsyn till olika keramiska lager. Ett idealt material skulle innebära en keram med modifierad CTE, ett ”tie layer” för kohesion, stängd porositet och termisk och kemisk inert i överhettarmiljö.
iv VÄRMEFORSK
Executive Summary This project has been predominantly focused on surface interactions related to boiler deposits in superheater zones of Waste to Energy (WTE) and biomass boilers. We have reviewed the literature and mechanisms of deposit formation, the nature of the deposits, in particular the initial stages of deposition. It is not a classical thin boundary layer deposit, but can be 10-15 cm thick and with a very heterogeneous chemical and physical nature. The corrosion problems are complex, the fluid dynamics and thermodynamics are likewise. In order to simplify the understanding of the problem, we have reviewed the composition of fuels, and the composition of deposits relevant to the initial stages of deposition. While many factors are involved in this initial stage too, it is considered that the increase in roughness is brought about through condensation of volatile chlorides, in particular alkali chlorides which are abundant in biomass and waste streams fed in as fuel. While it is true that more complex chemistry may occur in the earliest stages, the importance of KCl seems clear from the literature as both a key early depositor on surfaces, and also a key source of aggressive chloride for corrosion.
From reviewing the literature, and from interpreting the literature with the aid of some calculations we can make some conclusions and recommendations, and possibly some new observations. In superheater zones with fireside temperatures between 800-1000°C, KCl can exist in the gaseous state until close to the cooler surfaces of the boiler tubes, at which time it can condense. This is well enough known that additives are used by plant operators to reduce alkali chloride concentrations in the vapour, giving rise to HCl(g) which is less harmful on account of their higher volatility and can be scrubbed.
Other interventions to reduce boiler tube damage by flue gas include the coating of steel boiler tubes with thin ceramic and ceramic composite coats. The role of these is to block corrosive chemical attack from the fireside without compromising heat transfer and to minimize deposits. One great challenge with ceramic coatings of ferrous and nonferrous alloy boiler tubes is differential thermal expansion leading to cracks or thermal shock leading to wholesale delamination and other failure modes. Various ceramics can be tailored to minimize coefficient of thermal expansion mismatch and yet these are not necessarily of the best chemical composition to remain inert with respect to the fireside chemistry.
Inevitably, thermal shock damage in the surface coatings on boiler tubes can be reduced by spreading stresses, for example, using bond layers or through nanocompositing and by maintaining a certain level of porosity that can accommodate volume changes. Cycling of tube temperatures during operations can affect the interfaces in these materials and inevitably, even in the most thermally shock resistant coatings, nano- and microfractures will play the role of stress distributors.
In making ceramic coatings compatible with steels there is invariably porosity or nano and microfractures for gaseous KCl to diffuse into. These will offer a barrier only as long as these fracture systems remain disconnected, and this is probably not the case in
v VÄRMEFORSK
today’s materials. Secondly the porosity formed during slip casting and green body firing of sol gel ceramics will remain to also be a conduit for KCl gas.
Pore sizes in thermally shock-resistant ceramic coats are likely to be in the nanometric range. The Kelvin equation tells us that KCl will condense into these pores below the saturation pressure of KCl gas in bulk, so KCl liquid is stabilized even at high temperatures. Furthermore these liquids can be stabilized by small pore diameters due to the freezing point depression of liquids in small pores. These liquids can then migrate to the ceramic-alloy interface and cause harmful corrosion. Even worse are the eutectic melts such as can form from KCl and FeCl2 which can be stable in the boundary layer down to almost 200°C.
In order to investigate the role of physical adhesion mechanisms such as van der Waals forces in the formation of boiler deposits this review focuses on the initial layer of deposition that leads to further build up. In this respect, we concentrate on the role played by alkali chloride compounds, primarily KCl. Surface forces are reviewed in considerations of the above challenges, including effects of surface roughness on adhesion forces, the effect of interfacial wetting by fireside liquids and interparticle forces operating in the fireside flue gases.
The conclusion is that the best method to minimize corrosion, maintain fewer build ups and to maintain the life of a ceramic coating is to use a thermally shock resistant nanocomposite ceramic with a bond layer to the boiler tube alloy while at the same time using additives to lower the alkali chloride load in the flue gas.
Further work needs to be done on the exact surface chemistry of KCl liquid vapour solid equilibrium in respect to interactions with silica bed-material in fluidised bed boilers and also the general surface and capillary science of cracks and microfractures with respect to various condensates in the ceramic layers. An ideal material would include a ceramic that is composed for CTE matching, has a bond layer for cohesion and has closed porosity and thermal and chemical inertness to a range of fireside chemistries.
vi VÄRMEFORSK
Table of contents
1 INTRODUCTION ...... 1 1.1 BACKGROUND ...... 1 1.2 AIM AND TARGET GROUP ...... 1 2 BOILERS AND FUELS ...... 2 2.1 BOILERS ...... 2 2.1.1 Grate-fired ...... 2 2.1.2 Fluidised beds ...... 2 2.1.3 Pulverised combustion ...... 2 2.2 FUELS ...... 2 2.2.1 Biomass ...... 3 2.2.2 Waste ...... 4 2.3 COMBUSTION ...... 4 2.3.1 Fundamental combustion phases ...... 4 2.3.2 Release of alkali, chlorine and sulphur ...... 5 3 FORMATION OF DEPOSITS ...... 6 3.1 FLUE GAS COMPOSITION ...... 6 3.1.1 Ash elements ...... 6 3.1.2 Chlorine ...... 6 3.1.3 Sulphur ...... 6 3.2 MECHANISMS ...... 6 3.2.1 Inertial impaction ...... 6 3.2.2 Diffusion ...... 7 3.2.3 Thermophoresis ...... 7 3.2.4 Condensation ...... 7 3.2.5 Chemical reaction ...... 7 4 TYPICAL TUBING MATERIALS AND CORROSION ...... 8 5 SURFACE CHEMICAL PROPERTIES OF A DEPOSIT ...... 9 5.1 LIQUID DEPOSITS ...... 9 5.1.1 Condensation ...... 9 5.1.2 Chemical reaction ...... 10 5.2 SOLID DEPOSITS ...... 11 5.3 EXAMPLES OF SOLID DEPOSITS FOUND ON SUPERHEATERS ...... 11 6 COATING MATERIALS...... 13 6.1 COATING MATERIALS FOR PRESENT AND FUTURE BOILER TUBES ...... 14 6.2 AN EXAMPLE OF A COMMERCIAL NANOCOMPOSITE COATING ...... 15 6.2.1 Review of a boiler tube coating patent ...... 16 6.2.2 Comments on preparing boiler tubes for coating ...... 20 6.2.3 Other commercial systems for coating boiler tubes ...... 20 7 ASPECTS OF SURFACE FORCES IN BOILERS ...... 22 7.1 INTRODUCTION ...... 22 7.1.1 Van der Waals and dispersion forces ...... 22 7.1.2 Relevance of optical data to boiler deposits ...... 23 7.1.3 Dielectric response functions ...... 24
vii VÄRMEFORSK
7.1.4 Optical data for KCl at high temperature ...... 26 7.1.5 Hamaker constants for alkali chlorides ...... 26 7.1.6 Dielectric response functions of ceramic and inorganic materials ...... 32 7.1.7 A comment on negative Hamaker constants ...... 34 7.2 ADHESION, CAPILLARY FORCES AND SMOOTH SURFACES ...... 35 7.2.1 Forces between surfaces ...... 35 7.2.2 Rough surfaces and adhesion ...... 37 7.2.3 Thin films of liquid alkali chlorides on surfaces ...... 38 7.2.4 Van der Waals forces and KCl aerosols ...... 39 7.2.5 Adsorption of alkali gas phases on clays ...... 40 7.2.6 Direct condensation of KCl(g) into surface pores ...... 41 7.2.7 Capillary forces and KCl liquids in pores ...... 42 8 CONCLUSIONS ...... 48 9 RECOMMENDATIONS AND USE ...... 50 10 SUGGESTIONS FOR FURTHER RESEARCH ...... 51 11 LITERATURE REFERENCES ...... 52
APPENDICES
viii VÄRMEFORSK
1 Introduction
1.1 Background The high content of alkali (sodium and potassium) and chlorine in waste and biomass leads to operational and maintenance problems through the formation of deposits and subsequent corrosion of the heat transfer surfaces, particularly on superheaters. Operation is in itself also a problem because deposits affect the heat transfer. To remove deposits, soot blowing is used, but it leads to wear on the heat exchange surfaces and eventually also to increased material loss of the tubes. Thus, it is desirable to avoid the formation of deposits in the first place.
Formation of deposits during combustion and corrosion of heat transfer surfaces has been widely studied in several Värmeforsk projects. Only one Värmeforsk study about smooth tubing surfaces, i.e. surfaces with a non-stick effect, has been performed [1]. There are more experiences in this field, and a need to compile them. The purpose of this work is to review state of the art smooth surface research and technology for fouling resistant surfaces in boiler tubes, with emphasis on superheaters. To achieve this, we combine expertise in both surfaces and boilers. Specifically we utilize expertise in coatings, surface chemistry, nanotechnology and surface adhesion presently at YKI (Ytkemiska Institutet, part of the SP group). We combine this with expertise in alkali chemistry and deposit formation in boilers and formation and emission of ash particles relevant to boilers currently at SP Technical Research Institute of Sweden.
1.2 Aim and target group The aim of this work is to present a comprehensive state-of-the-art review of deposit- resistant smooth surfaces for boilers. It has a special focus of superheaters. The target group of this work are owners of heat and power plants, designers in this field, those responsible for operation and maintenance, and researchers.
1 VÄRMEFORSK
2 Boilers and fuels
2.1 Boilers Different techniques are used to combust solid fuels. They differ mainly with respect to the furnace; i.e. where most of the conversion of the fuel takes place. The major techniques are grate-fired, fluidised bed and pulverised fuel firing. All techniques may be used for power production, which requires water to be vaporized and superheated. The steam tubes in which this takes place are called superheaters. They are usually placed downstream of the furnace and are therefore exposed to the stream of flue gas.
2.1.1 Grate-fired In a grate-fired boiler, the fuel is combusted on a grate at the bottom of the furnace. Air is supplied through nozzles in the grate. Usually, the grate is moving so that the fuel enters from one side of the furnace and is transported during conversion. Finally, bottom ash is removed from the other side of the grate. The grate-fired technique is relatively insensitive to the fuel quality as for the size of fuel particles and moisture. However, the mixing of fuel and air may become uneven.
2.1.2 Fluidised beds A fluidised bed consists of fluidised sand in a reactor. The sand is usually quartz-based. Air for fluidisation and combustion is supplied through nozzles at the bottom of the furnace. Above a certain air velocity the sand becomes fluidised, which means that it attains the characteristics of a liquid. The contact between sand and fuel implies high heat transfer, and the turbulent characteristics of the fluidised bed provide good mixing. The high heat transfer to the fuel particles allows the bed temperature to be relatively low, which implies lower formation of thermal NOx compared with other combustion techniques. In a circulating fluidised bed the air velocity is so high that the sand particles are lifted by the gas stream. They are later separated from the gas in a cyclone and returned to the furnace. Most solid fuels can be fired in a fluidised bed as long as the fuel particles are not so large that they are not fluidized. Some alkali-rich fuels may cause the particles of the bed sand to agglomerate and eventually cause de-fluidisation.
2.1.3 Pulverised combustion In pulverised fuel firing, the fuel is milled before combustion. The small fuel particles are introduced together with air to the furnace by burners on the furnace wall. The fuel has to be dry and possible to mill. This technique is suitable for coal and some pretreated biomass (e.g. torrified biomass).
2.2 Fuels This report focuses on solid biomass and has a Swedish focus. Consequently it has the fuels biomass and waste as starting points. Typically, fuel related problems like formation of deposits vary from minor for wood to more challenging for straw, agricultural residues and waste.
2 VÄRMEFORSK
2.2.1 Biomass Biomass includes all kinds of materials that were directly or indirectly formed through photosynthesis, not too long ago. The main components of plant matter include cellulose, hemicelluloses and lignin. The concentration of each component in plant matter depends on species, type of plant tissue, stage of growth, and growing conditions. Three main classes of biomass are:
1. Wood and woody materials 2. Herbaceous and other annual growth materials 3. Agricultural wastes and residues
In Table 1 compositions and heating values for a number of biomass fuels are presented. For comparison waste fuels are included as well. With respect to the formation of deposits, the ash content and its composition are crucial, but also the concentrations of sulphur and chlorine are important. The least problematic biomass wood pellets contain less than 1 % ash and negligible amounts of sulphur and chlorine, whereas, wheat straw is known for causing problematic deposits because of its high content of chlorine and ash.
Table 1. Fuel analyses of a number of fuels. Composition in mass-% of dry fuel, and low heating value (LHV) in MJ/kg dry fuel. MSW = Municipal Solid Waste. IW = Industrial Waste. Tabell 1. Bränsleanalyser för ett antal bränslen. Sammansättning i mass-% torrt bränsle, och lägre värmevärdet (LHV) i MJ/kg torrt bränsle. MSW = Municipal Solid Waste. IW = Industrial Waste.
Fuel a C H O N S Cl Ash LHV
Wood pelletsb 50.6 6.0 43.1 0.08 <0.01 <0.01 0.3 19.12 Willow woodb 49.9 5.9 41.8 0.61 0.07 <0.01 1.71 18.40
Wheat strawc 44.9 5.5 41.8 0.44 0.16 0.23 7.02 16.83 Switchgrassc 46.7 5.8 37.4 0.77 0.19 0.19 8.97 16.89
Almond shellsd 49.3 6.0 40.6 0.76 0.04 <0.01 3.29 18.28 Oat graind 47.6 6.6 40 2.2 0.19 0.05 3.1 18.49
Mixed papere 48.0 6.6 36.8 0.14 0.07 - 8.33 19.45 60% MSW/40% IWe 44.3 5.6 26.4 1.1 0.4 0.5 21.7 15.86 a Fuel analyses are from reference [2] cited by reference [3], except for wood pellets [4] oat [5] and 60%MSW/40% IW [6] bWoody fuels c Herbaceous and other annual growth materials dAgricultural wastes and residues eWaste
3 VÄRMEFORSK
2.2.2 Waste Waste fuels are more heterogeneous than biomass. They also vary with respect to chemical content depending of their origin, e.g. plastics, paper and wood. Compared to biomass, waste often has higher ash content (Table 1) leading to severe formation of deposits on superheater tubes.
2.3 Combustion A fuel particle that enters the furnace undergoes the fundamental combustion phases drying, devolatilisation, and char combustion. Depending on size and shape, these phases can take place at the same time in different positions in a particle. In the following description it is assumed that the particle is so small that the processes are consecutive. During these phases, alkali, chlorine, sulphur and other species important for deposit formation are released to gaseous phase.
2.3.1 Fundamental combustion phases Drying starts as soon as the particle gets heated. The temperature increases to about 100 °C where it remains until the drying is complete.
Devolatilisation starts as the temperature reaches about 200 °C. The thermal energy causes chemical bonds to break and thus volatile compounds are released. In biomass, the first molecule to break is hemicellulose, and thereafter cellulose. Lignin is devolatilised over a wider temperature range, but at 500 °C, most of the devolatilisation reactions have been completed. Waste consists of many different materials. The major difference between waste and biomass is the presence of plastics, but a major part of those also devolatilise in the range 200-500 °C. The devolatilisation products are lumped into gas and tar. Tar consists of condensable organic compounds of different molecular size. The condensation temperatures differ between these compounds. Most of them condense below 300 °C, but some at higher temperatures. The gases are compounds that are not condensable at room temperature. They comprise CO, CO2, CH4, H2 and others. As the tar and gas meet oxygen, an oxidation process takes place. The radiated heat is visible as a flame. The heat sustains the devolatilisation process. The phenomenon usually referred to as "fire" is actually devolatilisation of e.g. wood, and the heat for the devolatilisation is provided by the flames.
Char combustion begins as the rate of devolatilisation decreases, since then oxygen can reach the char surface. The char is the remaining solid of the fuel after devolatilisation, and it has a high content of carbon. When the oxygen meets the carbon, the reaction produces CO2, and the heat released in the process results in a visible temperature rise of the char surface; usually referred to as glow. The solid remains after char combustion are ash. Some of the ash forms particles small enough to be carried away by the gas flow and is therefore called fly ash. The rest is bottom ash. A small fraction of the ash is also transferred to gaseous form.
4 VÄRMEFORSK
2.3.2 Release of alkali, chlorine and sulphur The release of alkali, chlorine and sulphur to the gaseous phase occurs by devolatilisation and/or evaporation. A small fraction of the alkali and a substantial fraction of the sulphur in the fuel is organically bound, which means that alkali and sulphur are released during the devolatilisation. Evaporation of inorganic compounds increases with the temperature of the fuel because of the rising vapour pressure. Chlorine is released as HCl in the same temperature range as the devolatilisation. Additional chlorine is released above 600 °C as the vapour pressure of alkali chlorides increases. At higher temperatures alkali sulphates may be evaporated. To summarise, alkali chlorine and sulphur are expected to be released in two temperature ranges. HCl, SO2 and a small amount of alkali are released at typical devolatilisation temperatures (200-500 °C), and alkali chlorides and sulphates evaporate during char combustion (>600 °C).
In the literature it has been found that the release of chlorine during thermal conversion of straw appears in two temperature ranges [7]; about 60% was released between 200 and 400 °C, i.e. during devolatilisation, and the rest between 700 and 900 °C. Likewise, SO2 and alkali have been found to be released during devolatilisation and at higher temperatures [8, 9]. However, the release of alkali during the devolatilisation constitutes only a small fraction of the total release [10,11].
5 VÄRMEFORSK
3 Formation of deposits The main content of severe deposits is typically alkali chloride, i.e. potassium chloride and sodium chloride. However, firing waste may also lead to a more complex deposit composition, e.g. lead and zinc may be crucial. The formation of deposits depends on the flue gas composition and by what mechanisms some of its content may deposit on steel tubes.
3.1 Flue gas composition The flue gas consists of a mix of gaseous compounds and particulate material. Most important for formation of deposits are ash elements, chlorine and sulphur.
3.1.1 Ash elements During combustion, ash elements are released from the fuel (c.f. 2.3.2). Alkali, i.e. potassium and sodium, is the major problem. In biomass, potassium dominates, and in waste both potassium and sodium are important. Potassium and sodium may cause low- melting compounds, which lead to agglomeration of the fuel bed or formation of deposits on heat transfer surfaces. The detailed ash reactions in a specific boiler are mainly determined by temperature, the ash’s composition and concentration, and possible bed sand. For example, the release of alkali and subsequent formation of alkali chloride increases with temperature. By decreasing the temperature in the fuel bed one can decrease the formation of deposits in the superheater region [12]. The released ash elements often deposit on superheater surfaces, and if they form alkali compounds with low melting points there is a risk for corrosion, especially in cases with high material temperatures.
3.1.2 Chlorine When chlorine is present, the released alkali easily forms alkali chloride. Therefore, high chlorine content in the fuel is problematic with respect to formation of deposits when the concentration of alkali in the fuel is high. High content of chlorine also means a high formation of acid HCl, but that is another problem.
3.1.3 Sulphur When the sulphur concentration in the flue gas is high alkali sulphates may be formed. Sulphates are less sticky and less corrosive than chlorides so in this sense, sulphur has a positive effect on deposits. It is well known that problems with superheater deposits can be reduced by adding sulphur to the process [13, 14].
3.2 Mechanisms
3.2.1 Inertial impaction Inertial impaction refers to deposition of particles. Particles carried by the flue gas generally follow the gas stream. However, when the gas reaches a steam tube it has to make a sharp bend. Some of the particles may be heavy enough (have enough inertia) to
6 VÄRMEFORSK
continue in the previous direction, i.e. not following the gas stream. This means that they are liable to hit the surface of the tube. Depending on the structure of the surface or the particle, it can then become stuck on the surface, thus building up a deposit. Inertial impaction is especially important for deposition on the windward side of a tube.
3.2.2 Diffusion Diffusion refers to the deposition of particles (mainly size < 1 µm) and gaseous molecules. It occurs if there is a concentration gradient with respect to particles or the molecule of interest. Then, for statistical reasons, a net movement of particles/molecules occurs from higher to lower concentration. If the gradient is present close to the surface of the steel tube, the molecules can condense on, or chemically react with, the surface and thus build up a deposit.
3.2.3 Thermophoresis Thermophoresis refers to the deposition of small particles. In a sharp temperature gradient of the flue gas, the side of a particle that faces the warmer gas will experience harder and more frequent collisions with the flue gas molecules than the opposite side. This results in a force that moves the particle from warm towards colder flue gas. Because the surface of the steel tubes is typically at 450 °C and the flue gas may be 600-800 °C, there is a temperature gradient that can move particles to the surface.
3.2.4 Condensation Condensation is a phase transition from gaseous to liquid form. It occurs when a gas reaches a partial pressure above the saturation pressure, which is temperature dependent, why cooling of the gas may initiate condensation. Condensation may occur by homogeneous nucleation in the bulk of the gas, but more likely on a surface. Surfaces in the flue gas are provided by the particles carried by the flue gas. These particles then get a partly liquid surface, which increases the probability that it will get stuck if it reaches the surface of the steel tube e.g. by inertial impaction or thermophoresis. Condensation may also occur directly on the surface which is usually much colder than the gas. In such a case, if diffusion is the transport process, the condensation maintains the concentration gradient in the gas phase, and is thereby a condition for deposition by diffusion.
3.2.5 Chemical reaction Chemical reactions may occur between species in the flue gas and the steel surface or deposits on the steel surface. If the product is liquid or solid it builds up the deposit. Chemical reactions can also take place on particles in the flue gas. If the product is liquid it can enhance deposition in the same way as condensation on particles.
7 VÄRMEFORSK
4 Typical tubing materials and corrosion Superheater corrosion usually occurs due to high material temperatures in combination with alkali compounds with low melting points and a reducing atmosphere. Different tubing materials are used in different positions of a boiler, depending of the need of corrosion resistance, with respect to temperature and surrounding flue gas. The steel grades used in a boiler can be sorted into five groups, see table 2 [15]. Most simple is carbon steel/low-alloyed steels, which contains less than 3 % Cr. An example of this is the commonly used superheater material 15Mo3. Low-alloyed steel forms iron oxide during combustion. The other steel groups contain higher amounts of chromium, which leads to formation of a continuous layer of a chromium-rich oxide that protects against corrosion. In boilers firing biomass or waste the most commonly used tubing materials for superheaters are austenitic stainless steels and high-alloyed Fe-based steels [15]. These steels provide good value for money with respect to corrosion resistance. Most resistant against corrosion are the alloys with highest Cr/Fe value, i.e. high alloyed Ni- based alloys. However, such materials are also most expensive.
Table 2. Common alloys used for high temperature applications (after [15]). Tabell 2. Vanliga legeringar som används vid högtemperaturapplikationer (efter [15]).
Steel/alloy Fe Cr Ni Mn Si Mo Cr/Fe Carbon steels/Low alloyed steels 15Mo3 99 - - 0.6 - 0.3 0
13CrMo44 98 1 - - - 0.5 0.01 T22 96 2 - 0.5 - 1 0.02 Ferritic chromia formers P91 89 9.5 0.1 0.5 0.3 0.9 0.11 X20 85 12 0.5 1 - 1 0.14 433 77 22 0.1 0.6 0.2 - 0.29 Austenitic stainless steels Esshete 1250 67 15 9 6 0.5 1 0.22 316L 67 17 12 1 0.4 2.5 0.25 347H 68 17 11 1.7 0.6 0.3 0.25 304L 68 19 10 1.4 0.6 0.5 0.28 253MA 66 21 11 0.4 1.5 - 0.32 High alloyed Fe-based steels 310 51 26 18 1.6 0.9 - 0.51 AC66 40 28 32 - - - 0.70 353MA 36 26 33 1.4 2.4 - 0.72 Sanicro 28 35 27 31 2 0.7 3.5 0.77 High alloyed Ni-based alloys Inconel 690 9 30 60 - - - 3.33 Inconel 625 5 22 58 0.5 0.5 9 4.40 Sanicro 63 2 22 65 - - 9 11.00
8 VÄRMEFORSK
5 Surface chemical properties of a deposit The present work concerns the initial build-up of deposits on the surface of a steel tube and how it can be minimized through materials engineering. It can be assumed that liquid compounds that reach the surface stick to the surface. For solid substances, the attractive forces between the molecule and the surface have to be compared with other forces affecting the molecule.
5.1 Liquid deposits The liquid substances can reach the surface by diffusion coupled with condensation or a chemical reaction. Liquid substances can also be formed on particles in the flue gas and be transported to the surface by impaction, diffusion or thermophoresis.
5.1.1 Condensation The process of condensation implies, by definition, that a liquid is formed on the surface from a gaseous species. To judge whether condensation of a certain substance will occur, the concentration/partial pressure in the gas phase and the temperature of the surface have to be known. This is compared with the saturation pressure at that temperature. If the partial pressure is above the saturation pressure, condensation can occur. As will be seen later, condensation can occur on surfaces when the partial pressure is lower than the saturation pressure at which the liquid condenses under certain conditions of surface geometry. This sub-saturation condensation can occur if the condensed liquid normally wets the surface, and if the surface has nanoscale roughness or accessible nanoscale porosity. We will return to this in chapter 7.
As an example for the case of smooth surfaces, the saturation concentration of gaseous KCl as a function of temperature is given in Figure 1. It shows that for condensation to occur on a smooth surface at say 800 °C, the concentration must reach around 2000 ppm, which is an unlikely high concentration. If, instead, the concentration is around 100 ppm at 800 °C, condensation should occur if the gas were cooled to around 700 °C. Considering that superheater tubes seldom are warmer than 500 °C, it can be concluded that there would be risk for condensation as the flue gas approaches the tube surface.
The flue gas is convectively cooled by passing the steel tubes. Condensation may then occur on particles in the flue gas. These particles may thereafter come in contact with the surface of a steel tube and stick there.
9 VÄRMEFORSK
Figure 1. Saturation concentration of KCl versus temperature [16, 17]. Figur 1. Mättnadskoncentration av KCl mot temperatur [16, 17]
5.1.2 Chemical reaction A liquid deposit may result from the chemical reaction between a gaseous species and a solid species on the surface, or between a gaseous species and the surface itself. The prerequisite is that the product is liquid. For example, the reaction
G + S → L, where G is a gaseous species, S is a solid species on the surface, and L is the liquid product, shows such a process. In practice, however, it seems unlikely that this process would be of great importance, since it requires firstly that a heterogeneous reaction take place, and secondly that the product thereof be liquid at the temperature of the surface.
10 VÄRMEFORSK
5.2 Solid deposits Solid deposits may build up simply from adhesion of particles in the flue gas. The particulate matter may consist of pure mineral species, soot, oxides or complex aggregates of various species. The size range of particles may vary from nm to microns. The adsorption of particles to the surface governed by a force balance between attractive forces including van der Waals forces, and electrostatic interactions and repulsive forces including viscous drag of the gas around the particles. The solid deposit therefore reflects to some extent the distribution of species in the flue gas but more accurately which type of particulate matter that has highest affinity for the surface.
5.3 Examples of solid deposits found on superheaters Jensen [18] examined deposits from straw-based biomass boilers in Denmark. The two plants were Ensted and Masnedø of 100MWth and 33MWth, respectively. The Ensted superheater runs slightly hotter than the Masnedø superheater, with final steam temperatures of 542 °C and 520°C respectively. Deposits at Masnedø were 2-15 cm thick, while at Ensted they had a maximum thickness of a few centimetres. Both were comprised of three layers, with one relatively thick intermediate layer.
The thick middle layer of deposits at Masnedø was formed when molten KCl reacted with a calcium-rich silicate ash-derived phase. The reaction formed a chlorine-depleted layer due to conversion of chlorine to gas. The inner most layer of the deposit contained many sublayers of mainly iron oxide with KCl and K2SO4. The Ensted intermediate layer comprised KCl with melt textures, with these having inclusions of Ca- and Si-rich particles. The innermost layer at Ensted was an iron oxide layer next to a potassium sulfate layer. Jensen [18] also found that the mature deposits were different in composition compared with that of experimental probes exposed for a short time, suggesting that reactions continue after deposition. The mature deposits form dense layers next to the boiler tube surfaces that are rich in KCl and K2SO4. The same authors postulate that the innermost layer continues to expand by condensation of gaseous KCl that has diffused into the preformed deposits, giving rise to a more and more dense and K-rich deposit as time proceeds. They noted that in the case of straw, potassium was not present in the sulfate form, which implies that the potassium sulfate deposits seen on the boiler tubes must have come from the reaction with sulfurous gases, such as may occur by catalytic oxidation of SO2 on iron oxide. The oxidized SO2 species could then react with KCl to yield K2SO4 plus chlorine gas. Their thermodynamic calculations also indicate that in the case of excess chloride, the potassium species will be gaseous KCl up to about 1200°C, and if sulfate is present in the gaseous phase above about 780°C, potassium sulfate will be thermodynamically more stable than the chloride.
KCl is considered far more corrosive than K2SO4 – the reason why studies of additives that favour the formation of sulfates have been performed. These studies include research on additions of sulfur sources and additions of clay minerals such as kaolinite (see Wu et al. (2011) and references therein) [19]. These interventions are Al, Si, S, P and Ca-based and lead to predominantly more potassium sulfate and less alkali chlorides depositing on boiler surfaces. The chlorine in alkali chlorides is converted to gaseous HCl, which can escape deposition. These interventions can reduce NOx output
11 VÄRMEFORSK
and decrease corrosion rates, but can also increase sulfurous gas outputs. In any case the use of additives will not be considered further here, other than to say that if coatings are used on boiler tubes then the chemistry of the flue gas plus additives should be taken into account as they will have different reaction pathways compared with burning of the raw fuel.
In the case study of Liu et al. [20], the ternary mixture KCl- K2SO4- Na2SO4 as components of biomass-fired boiler deposits was examined. The samples were collected from various super-heater zones of a 30 MW boiler. The deposits discussed here are mainly from the secondary superheater zone (SH2) where flue gas temperatures were intermediate at 550-750°C and steam temperatures of 430-470°C. The fuel was cotton stalk. The higher temperature zones formed deposits rich in SiO2 and Ca-(Fe,Mg)Si2O6 pyroxenes. Any salts formed at this temperature were probably evaporated and carried to lower temperature surfaces. In the SH2 zone, the initial layers were < 2mm thick oxide scales, as determined by elemental analysis. Iron oxide was present as Fe2O3 as determined using X-Ray diffraction (XRD).
12 VÄRMEFORSK
6 Coating materials
It is apparent that the problem of producing a fuel-independent coating technology for boiler alloys is a problem of high-temperature chemistry and physics, and low- temperature solution nanochemistry and corrosion chemistry. Ideal anti-soil coatings must resist the creep, thermal expansion and oxidation of boiler tubes in their normal operations, and the chemical attacks. Nor should they offer additional thermal resistance. In addition they should be resistant to inertial impacts that erode, pit crack and generally fatigue the surface coating in ways that lead to non-smooth surfaces that can lead to formation of deposits. They should of course be affordable and practical to apply.
The interface and the elastic/plastic behaviour of a high performance coating will provide a natural gradient over which adverse solid-state interfacial tensions between various materials can be alleviated allowing well-bonded, semi-elastic ceramic layers with extremely smooth surfaces and high temperature chemical resistance. Various approaches to achieving highly adherent coatings include the use of intermediate layers (e.g., bond layers, solid solutions) at the metal/metal oxide ceramic interface (for both alloy and ceramic coatings). For example these can be formed during sintering of ceramics or by thermal spray of appropriate compositions [21-25]. The mechanics of stressed ceramics is relevant to these approaches has been examined in Sweden, for example by Odén et al. [26].
Next generation boiler tubes will need to be designed with extreme care considering the plethora of interfaces in these highly engineered systems, involving nano-scale structures with high specific surface areas. The primary purpose of this section is to point to other reviews on the subject of coating compositions, nano-compositing and coating methods, and to identify some illustrative cases from the academic literature and industrial patent literature.
13 VÄRMEFORSK
6.1 Coating materials for present and future boiler tubes This section briefly reviews coating materials for protecting boiler tubes from formation of deposits in combustion facilities. It briefly introduces several classes of oxides, ceramics and other crystal classes found in the literature survey and then gives some examples of present and future potential materials from academic and commercial sources.
The definition of the term “refractory” is key to most materials relevant here. Formally it means that the material does not evaporate readily. The elements that form refractory oxides generally are found in left corner of the d-block transition metals on the periodic table (Figure 2).
Figure 2. Basic periodic table of the elements [27]. Figur 2. Periodiska systemet [27].
Elements forming chemically resistant refractory oxides are from the left side of the d- group transition metals and include Ti, Y, Zr, Nb, Hf, Ta, W and also from the p-group including Al. Of these alumina is the cheapest and it appears in several patented nanocomposites for boiler tube coatings reported below, and is probably the most studied of the refractory oxides. Titania in the high temperature form is rutile and is also found in several recent patents for boiler tube coatings. Oxides of Y (yttria) are extremely expensive, but are extremely useful because of their chemical inertness at high temperature and extreme resistance to temperature, so much that it is often used in foundries as a ceramic for casting superalloys for jet engine turbine blades and other performance materials. Yttria, amongst other additives is also useful as a stabilizer of the tetragonal phase of ZrO2. Zirconia is the last of the relatively affordable elemental refractory oxides, and for this reason it is also to be found in the nanocomposites disclosed in several patent applications. Nb, Hf, Ta and W oxides are only considered likely to be practically useful when used as minor additives, or as solid solutions components in other oxides such as rutile, for which Ta and W oxides share the same crystal structure.
In addition to the elemental oxides, more complex crystalline refractory materials appear to be useful enough for companies to be actively patenting these. In these
14 VÄRMEFORSK
materials, other elements become useful such as B, C, N, Si, Sn, alkaline earth metals, several transition metals (Cr, Fe etc.), and more, but usually (for complex oxides) when Al, Ti, Y and Zr occur with them in the structure. They are often doped with other elements to tune their properties and form equilibrium compounds with other refractory materials in their multi-elemental phase diagrams. They are also candidates for coating of SiC as discussed below. Examples often include layers or composite layers of various spinels (CuAl2O4, MgCr2O4, yttria-doped-MgAl2O4, (Fe,Zn)Cr2O4), doped-rutiles, yttrium aluminium garnets (YAG), pyrochlores, perovskites (YCrO3, CaTiO3), mullite . (nAl2O3 mSiO2), various zirconates and titanates (Al2TiO5/Al2O3 composites, ZrTiO4, BaZrO3, Al2TiO5 and CaTiO3), Gd2Zr2O7, V2O5, silicon carbides, boron carbides, silicon nitrides, aluminium nitride/silicon carbide composites and C-CSi composites [28-30].
Boss [30] provides a comprehensive review of potential composite materials for SiC. They conclude that no other material beside SiC could provide the combination of resistance to thermal creep, thermal shock and oxidation necessary for constructing externally-fired combined cycle (EFCC) power systems. For example, Si3N4 was considered to have too low a coefficient of thermal expansion (CTE). They also concluded that corrosion by coal combustion products severely degraded SiC so their study was aimed at coating SiC. A large number of materials and their properties were considered.
6.2 An example of a commercial nanocomposite coating A nanocomposite material called Nanocomp PP has been tested on the metal tube surfaces on boilers tube surfaces in Sweden [1]. Nanocomp PP, according to an early article in German [31], is a nanocomposite of nanoparticulate, hexagonal boron nitride (“white graphite”) and ceramic nanoparticles nominally zirconia (judging by their patent application – [21]), but could also be Al2O3, AlO(OH), ZrO2, Y-ZrO2, TiO2, Fe2O3 and/or SnO2. While it stayed adherent to the boiler tubes, it provided significant benefits in terms of corrosion and soil resistance [1]. However the retention of the coating was significantly impacted in a relatively short time.
The patent application for Nanocomp PP technology from ItN Nanovation AG is for coating surfaces in Waste to Energy (WTE) and other boiler based plants. The patent number is EP1386983B2 (in German) and refers specifically to boron carbide (BC) composites with a fairly brief teaching of its disclosure [21]. However it is possible that another set of patents is relevant, i.e., WO 03/93195 and US 20090253570 A1 [32], also from ItN Nanovation.
Even if this patent is not directly relevant to the studies reported in Värmeforsk report 1034 [1], a detailed look at this patent will help with understanding a wider range of materials and coatings for boiler tube application. Therefore a review of this patent is used to drive a broader understanding of ceramic coatings in general. When this review was written we were not aware of the US filing so translations from German are referred to below. In this mini-review, the patent technology is described along with simultaneous comments and observations of significant points.
15 VÄRMEFORSK
6.2.1 Review of a boiler tube coating patent US 20090253570 A1 (and WO 03/93195) [32] is a patent application, and describe the sintering of a composition comprising particles of component A with particles of component B and binder particles C. A and B form a spinel phase at the same time C is used as a binder during sintering at 900-1000°C, which is much lower than for most similar ceramics. This is standard operating procedure because of the choice of a eutectic mix, which has a lower temperature than otherwise achievable for the individual components. This lowering of the melting temperature through eutectic formation was first described in ItN Nanovations patent on boron nitride nanocomposites for boiler tubes, and was the first topic in the body of the spinel-based patent applications being reviewed here, such is its centrality to the invention.
“A composition is provided for producing high-temperature-stable ceramic …layers, coatings and moulded bodies and particles of an inorganic component A and particles comprising an inorganic component B. This may together form a eutectic mixture and reacted at … at least partially with each other, where it is formed at least a ternary chemical compound (i.e. a combination of 3 elements), and in particular the spinel type.”
Coating ingredients, A and B Spinels of the disclosed composition are of the type AB2X4. The A component being oxides of the transition metal, rare earth metal and p-group metals of the periodic table, with Cu, Fe, Co and Zn oxides are particularly preferred, in particular copper, and the B component chosen from trivalent metal oxides, most preferably Al2O3. Component A is added in proportions, in particular between 5-15 weight %, but as wide as 1-40%. Component B is added in proportions in particular in the range 70-90%, but as low as 50%. The preferred spinels so formed are copper aluminate spinels and chromium iron spinels. It is possible that for boiler tubes, iron chromium spinel nanocomposites might also be useful from a compositional similarity to the oxide layers of the tube alloy. This is not disclosed in their patent. They did disclose that the casting slip for forming a copper spinel/titania nanocomposite described below was applied to boiler tubes and tested for a very short time under nominal combustion in a power station boiler.
Binder ingredients, C (or excess B) Component C, the effective binder, can be one of A and B, and also can be chosen from at least one member from the group alumina, boehmite, zirconium oxide with yttrium- stabilized zirconia, chromium oxide, cerium oxide, iron oxide, silica, tin dioxide and more preferably titanium dioxide. The exact ingredients are chosen to respond best to the ternary spinel formed by A and B at the sintering temperature, in particular chrome iron spinel (Fe(Cr,Fe)2O4). As a side note, chromite spinels are used in the production of ferrochromium master alloys in arc furnaces. In addition, the binder can be the component B added in a quantity where the spinel forms by an excess of component B remains unreacted, and can form the binder phase with no addition of a third component. This is a particularly preferred embodiment of their invention. Component C is added according to the other ingredients in the range of 1-40% and in particular between 5-15 wt %.
16 VÄRMEFORSK
Other ingredients Other ingredients include suspending agents for the obvious dispersion stability of a rheological fluid with the right viscosity and other fluid properties for spreading and wetting the surface, but the exact amount is said to be not overly critical, thus lending some flexibility according to application. However, organics such as polyelectrolyte dispersants, often used for ceramics might be best used at lowest amounts to achieve homogenous suspensions to enable least final carbides and/or porosity. Fillers can also be used, such as mineral fibres, coarse particles etc. to add strength and other properties. It is said to be important that alkali and alkali earth elements are essentially absent from the mix as these can negatively interfere with formation and stability.
Example of a ceramic casting slip for coating boiler tubes To make up the casting slip nitric acid was made up at pH 2 in a bath and used as a suspension solution for the oxides. Next, α-Al2O3 (89.5 wt %) nano/microparticles (100 nm to 1 micron) were added to the nitric acid solution and homogenized. Nanoscale rutile was then added at a rate of 6.5 wt% with high energy stirring. Next, CuO is stirred in at 4 wt%. The solids content of the solution after homogenizing should be 80 wt%. The casting slip is then ready to work with and can be applied to a boiler tube. When this composition was sintered their example disclosed a phase mixture of 87% elongate nanorods of corundum (α-Al2O3), rutile (TiO2) and CuAl2O4-spinel. Formation of CuAlO2 was avoided by staying below 1000°C.
The nanoparticulate size of the ingredients should dramatically reduce equilibration times of the spinel formation. Jost et al. [32] indicate that the initial choice for the crystal habit of corundum is important for successful growth on a single crystal axis yielding fibres during sintering and for these same alumina crystals to seed the growth of copper oxide deposits in grain boundaries leading to a eutectic melting at the phase boundary at 90:10 alumina to copper oxide. The addition of nano-titania apparently aids the eutectic formation, possibly through nucleating low temperature melting.
Though the authors did not say, it is likely that in addition, nitric acid will also passivate the surface and will induce roughness enabling better physical interlocking with the ceramic. It is also likely that during acid treatment defects (in electrochemically reduced states) are oxidized in and on the existing porous oxide extreme outer layers. Dehydration probably also occurs, with iron hydroxides possibly converted back to oxides, making them denser phases and homogenizing the average oxide phase composition at the interface. In the case of iron chromium spinel coats, this will ensure a chemically cohesive and seamless transition to the iron chromium spinels in particular. A look at the literature on solid solutions for chromium iron oxides from the oxide layer of the tube alloy (plus or minus nickel) with these spinels will likely indicate the favourability of such solid solutions as bond layers between the steel and spinel coats.
Advantages of the nanocomposite solution coating methods described above include its high temperature corrosion resistance, relative inexpensiveness, the use of novel binding and sintering methods, the use of Cr and Fe oxides in the ceramic, and its impact resistance. The latter advantage is owed to the extremely high strength and hardness of the coat, having a bending modulus in the range 200-300 MPa, determined according to
17 VÄRMEFORSK
DIN ISO 60 672. The low chemical reactivity of the coat will slow the formation of its roughening during combustive operation, concomitantly slowing formation of nucleation sites for deposit growth. This decreases the potential for capillary related liquid wetting of the surface (see chapter 7) by KCl. Degradation of the smooth fluid dynamic aspects of the circular cross-section of the tube will also therefore be less of an issue as is the all important heat exchange performance. Of course these are all benefits assuming the patented composite lives up to its claims.
Given the importance of CuAl2O4 spinels in their patent disclosure, it is worthwile documenting its thermal expansion coefficient, for the record. The temperature dependence of CuAl2O4 spinels can be found very accurately described in O’Neill et al. [33]. They used neutron diffraction to investigate the temperature dependence of the cation distribution in the spinel. A graph of the unit cell parameter of spinel as a function of temperature is shown in Figure 3 below. As shown in figure 3b the linear -6 coefficient of thermal expansion of CuAl2O4 is around 8-10 *10 /ºC at temperatures typical of the boiler conditions. Steel has a linear CTE of around 13*10-6/ºC and Austenitic stainless steel has CTE in the range 10 to 17 *10-6/ºC . Typical boiler tube materials supplied by Sandvik for example have CTE in the range of 12-16 *10-6/ºC at temperatures of 400 ºC. The similarity in CTE of CuAl2O4 compared with steel and Austenitic stainless steel provides indication that this spinel should have favourable physical properties for tube coating. As a reminder most ceramics have very much lower CTE in the range 2-6 (e.g .SiC = 4*10-6/ºC ) which leads to thermal shock and cracking problems as mentioned earlier.
18 VÄRMEFORSK
a)
b) Figure 3. Thermal expansion of copper aluminium spinel. Figur 3. Termisk expansion av koppar-aluminium-spinell.
19 VÄRMEFORSK
6.2.2 Comments on preparing boiler tubes for coating Pre-cleaning of tubes prior to solution coating Another important consideration is good pre-cleaning. Mild sand, CO2 or water blasting of appropriately chosen and stored steel tubes should remove loose rust. A carefully selected non-ionic surfactant and co-solvent system should be used to remove any grease or other organic pollutant or soil films. Residues from the surfactants cannot be beneficial. So rinsing and careful choices of cleaning agent should be used.
Pre-passivation An important consideration for maximum adhesion of the ceramic layer is to have a high-strength, high density oxide layer to bond to at the steel surface. This could be comprised of fine-grained, interlocked crystals formed by classic nitric acid passivation. To induce a uniform, high-density oxide layer at the boiler tube surface, it could be pre- treated. Hydrogen build up should be carefully avoided during any pre-passivation as it could weaken the steel. If pre-passivation is not done, hydrogen bubbles could form from during various reactions of any exposed metal and cause defects during coating. Therefore it is recommended that native metal, accessible through surface pores, defects and surface cuts and scratches, be pre-passivated with nitric acid before coating.
Coating When the coating composition is to be prepared, high quality ceramic coating practices will help. These can be relatively cheap and repeatable and preferably complemented by use of colloidal science principles in preparing the surfaces. The skill is in fine-tuning the colloidal properties of the coating suspension (e.g. the slip) including the suspension homogeneity and rheology so it forms a uniform ceramic precursor (green body) layer. Most likely a spray coat and/or a vertical or horizontal dip coat is possible too. The description of these is beyond the scope of this report. Avoidance of alkali and alkali earths is important when preparing the boiler tube coatings. It would be recommended here that not only the coating composition be free of these during application, but also the steel surfaces be substantially free from these contaminants at the time of coating.
6.2.3 Other commercial systems for coating boiler tubes A number of commercial and institutional organizations have expertise, intellectual property and/or products in metal corrosion or ceramic coatings as solutions for materials protection in power plants. The most common ceramic coatings amongst these are alumina and fireclay-based (mainly kaolin as a precursor) which are more alkali resistant for biomass applications. Silicon carbide is also a key material in the burn zone as it has a larger thermal conductivity than most ceramics, and can thereby reduce the relative deposition rate for a particular fireside situation compared with a surface that cannot conduct heat away so readily. However it is not preferred for high alkali biomass fuels due to its reactivity as stated earlier. Most commercial manufacturers also try to keep the coating thickness thin in order to keep the tube surface-ceramic system close to isothermal. Judging by manufacturers comments, thin (perfect) coatings minimize differential thermal expansion to the point where it is not a problem. These coatings can be applied to tubes or to shells. Manufacturers are less prone to disclosing exact
20 VÄRMEFORSK
compositions and structures for obvious reasons. However as has been highlighted, nanostructured systems are more and more common and are expected to give higher all round performances per unit cost in current and future coatings. In use, whether coated or not, the manufacturers recommend that the operator keep the coatings as cool as possible for avoiding deposits under the given conditions. This can be for several reasons, the most important probably to reduce the “stickiness” of adhesive particles.
For particulars please see the reference list, noting some of these have been described above [21, 22, 31, 34-43]. It is also noted that thin (0.8 mm) corrosion coatings produced by arc spraying with bond strength of 25N.mm-2 are commercially available and have been tested in certain commercial labs [44].
A number of other studies exist that have examined boiler tubes, their coatings and their resistance to corrosion and dirt build up. The interested researcher is encouraged to go to these for valuable representative information pertaining to the current subject [1, 30, 45-47].
21 VÄRMEFORSK
7 Aspects of surface forces in boilers
7.1 Introduction
In the various surfaces of concern in current and future boiler tube materials, van der Waals forces are becoming more relevant. This is primarily because nano-scale structural engineering of interfaces and coatings is inherently dealing with higher surface areas. Nanoscale roughness, capillaries and nanoscale cracks and nanoparticles in the ceramic coating composites and in the aerosolized and condensing and abraded fireside atmosphere are heading quickly to ubiquity. Since this is the realm where van der Waals forces are operative (along with other forces), these are becoming more important to deal with. In this respect the reviewers recognized that in order to make progress in understanding the literature, some mathematical language relating to van der Waals (vdW) forces relevant to ceramics and liquids would have to be decoded and interpreted.
It was decided on the basis of known problems with alkali chlorides in biomass (and other) fuels that any new materials would have to be directly designed to deal with the major constituent of alkali chlorides, namely potassium chloride, and its interactions with as many possible constituents of present and future boiler tube surfaces as possible. In reviewing the literature, little detail was present on van der Waals interactions relevant to boilers. This next section deals with this. For those readers less interested in the mathematical details of calculation of dispersion forces, please skip to section 7.2. However more recent theoretical approaches suggest that current theories of interaction of macroscopic bodies, such as the modern dispersion theory used and reviewed here, are inadequate from many perspectives and need to be totally replaced with a new theory or theories [48]. These advances must be kept in mind while reading this review.
7.1.1 Van der Waals and dispersion forces The modern dispersion theory of interactions of macroscopic bodies relevant to colloids may be used to gain some understanding of the interactions of fireside particulates and gases with boiler tubes in burners. Dispersion forces are the component of the van der Waals force that arise from interactions between spontaneously formed dipoles and induced dipoles. The other components that involve permanent dipoles are the Keesom and Debye forces (dipole-dipole and dipole-induced dipole forces). London dispersion forces are mediated through exchange of electromagnetic radiation, and are therefore an optical phenomenon. A full analysis is not offered here, but the principles are given with a few simplified examples to pave the way for a more sophisticated approach, requiring further experimental data and probably sophisticated computer calculations.
Ninham and Parsegian [49] describe the (London) van der Waals force as “founded on the recognition that spontaneous, transient electric polarization can arise at a center due to motion of electrons, molecular distortion or molecular orientation. This polarization will act on the surrounding region to perturb spontaneous fluctuations elsewhere”.
22 VÄRMEFORSK
The modern theory of dispersion forces between macroscopic bodies is owed to Lifshitz [50], and has its roots in the presence of attractive forces between neutral atoms noted first by London [51] and its method based on the ideas of Casimir and Polder [52]. The interaction between macroscopic objects is regarded “as occurring through the medium of the fluctuating electromagnetic field which is always present in the interior of any absorbing medium, and also extends beyond its boundaries, - particularly in the form of travelling waves radiated by the body, partially in the form of standing waves which are damped exponentially as we move away from the surface of the body” [50]. Lifshitz’ theory overcame previous difficulties of calculating dispersion forces for materials other than dilute gases by including all many-body forces using a continuum approach, retaining all interaction frequencies and handling intervening dielectric media other than a vacuum. Importantly, Lifshitz showed that the information required for the calculation of surface forces were contained within the dielectric properties of the medium and interacting condensed matter.
In many cases, Ninham and Parsegian have shown that knowledge of the full optical spectrum of a material system is unnecessary for calculation of van der Waals forces using full Lifshitz theory [49, 53]. Instead, they demonstrated that a damped oscillator approach requiring only the average characteristic absorption frequencies in the UV and IR and indices of refraction at visible frequencies and the static dielectric constants were sufficient.
Hough and White [54] also discuss the Hamaker’s constant in relation to wetting of liquid films in an atmosphere against a surface. The condition for wetting is that the Hamaker constant for the system atmosphere/liquid film/solid should be negative. The Hamaker constant is calculated in the non-retarded regime (less than 5 nm separation) to give a reliable indication if a liquid film will wet a surface. If the film is wetting then one can apply the concepts of capillary condensation to liquid films in ceramic pores, cracks and corrosion pits, between ceramic nanocomposite components, and fireside solid and liquid aerosols such as KCl and its interactions with fluidized bed media and so on. The next section explores these interactions in more detail.
7.1.2 Relevance of optical data to boiler deposits It seems anti-intuitive that optical data, such as refractive indices and optical absorption coefficients can be relevant to boiler surfaces and deposits. But the London or dispersion part of van der Waals force which governs many interactions in the formation of boiler deposits, is mediated by fluctuations of the electromagnetic (optical) field in the vicinity of interacting bodies across all frequencies. This dynamic electromagnetic interaction involves light at various frequencies outside the visible range, and is dominated by the UV, and in decreasing significance, the IR and microwave [55]. Materials from metals to ionized gases are subject to dispersion forces. The numerous interactions where it dominates includes in the formation of colloidally stable rheological fluids for ceramic processing, through the formation of intergrain boundaries in ceramic nanocomposites and between metals and ceramics, and in the
23 VÄRMEFORSK
interactions of aerosols of fireside liquid and solid particles with each other and with ceramics and oxide coatings and layers on tubes. The mathematical constructions here were necessary in order to give an opinion on the significance of new material designs on dispersion force interactions, both from an inherent property point of view and from interaction during operations of boiler tubes. Below is a description on how full Lifshitz theory has been applied in the literature, and how it can be used to get new useful information on the interactions of fireside components with the ceramics and oxide coatings of the boiler tubes.
7.1.3 Dielectric response functions Non-retarded, full Lifshitz theory, involving frequency dependent spectral data subject to various Kramer-Kronig transforms can be used for construction of dielectric response functions of various materials [55]. The full, non-retarded Lifshitz theory can be applied to obtain the interaction parameter known from colloid science as the Hamaker constant, from which the London component of the vdW energy are derived.
In order to calculate these various frequency dependent dielectric response functions, Hough and White [54] showed that Parsegian and Ninham’s damped oscillator model could be written as follows:
(1)
where is the frequency dependent dielectric response function for a solid or liquid, CUV, CIR are the oscillator strengths in the UV and IR respectively, and ωUV, ωIR are the UV and IR oscillator damping frequencies, respectively.
Under the full non-retarded Lifshitz theory, calculation of Cauchy plots from refractive index can be made with limited spectral data, using a damped oscillator model (e.g., Hough and White) [54]. This requires the frequency dependent refractive index, to be known in a relatively slowly and smoothly changing part of the refractive index curve (in the visible range), and preferably in a frequency range where there is negligible absorbance. For most solid non-conductors the absorption strengths in the UV and IR can be derived from two equations. The first is the Cauchy equation, from which a Cauchy plot is constructed:
(2)
2 2 2 2 A linear plot of (n (ω)-1) versus (n (ω)-1)ω ) will have slope, 1/ ω UV and will intersect the ordinate axis at CUV. CIR can then be determined from:
CIR = ε(0) - CUV -1 (3) where ε(0) is the zero-frequency, static dielectric constant of the material.
24 VÄRMEFORSK
This gives a set of parameters ωUV, CUV, CIR, so the only missing parameter is the ωIR, the IR oscillator damping frequency, and is taken as the characteristic absorption frequency in the IR region. In practise for many materials this is a minor contributor to the overall dispersion energy. In the range of discrete Matsubara frequencies over which each contribution to the total dispersion energy is summed, only a few IR values occur, while there are hundreds in the UV part of the complex spectrum. Figure 4 below shows the first step in the procedure, using solid KCl as an example.
1.62 n KCl 1.6
1.58
1.56
1.54
1.52
refractive index, KCl index, refractive 1.5
1.48
1.46 200 400 600 800 1000 1200 1400 1600 1800 w avelength KCl Figure 4. Wavelength dependence of the real part of the complex refractive index for solid KCl [56]. Figur 4. Våglängdsberoende hos den reella delen av komplext brytningsindex för fast KCl.
The corresponding Cauchy plot is shown in Figure 5, constructed using the KCl refractive index data above. It shows excellent linearity in the range and allows a very accurate determination of the Cauchy parameters.
1.6 n^ 2-1 KCl
1.5
1.4
1.3 n^ 2-1 KCl
1.2
y = 1.1691 + 4.1175e-33x R= 0.99971 1.1 02 1031 4 1031 6 1031 8 1031 1 1032 (n^2-1)omega^2 KCl Figure 5. Cauchy plot for solid crystalline KCl. Figur 5. Cauchy-diagram för fast kristallin KCl
25 VÄRMEFORSK
Bergström [59], Faure et al.[62], Lee and Sigmund, Ackler et al., French et al., all go into some detail of calculations of Hamaker constants using this method. The authoritative works however are by Hough and White [54] and by Ninham and Parsegian [53]. Hough and White [54] and Bergström [59] calculated many of these Cauchy plots from reliable optical data, so further examples will not be given. In this review calculations have been included in order to better interpret literature on present and future boiler tube materials.
7.1.4 Optical data for KCl at high temperature Optical data for KCl at high temperature is rather limited. In the solid state, we can use the room temperature data, given the weak dependency of the Hamaker constant on temperature. For the liquid state, where the ionic liquid behaviour sets in, the frequency dependence will change, making a new set of data important to have, but none appears to be available after extensive searching. More information is given in Appendix 1.
7.1.5 Hamaker constants for alkali chlorides Hamaker constants are the scalar factor used to relate molecular interaction forces as a function of the separation between the bodies. Hamaker constants are calculated from imaginary frequency dependent dielectric response functions using full Lifshitz theory [50] using a limited damped optical oscillator model [49, 53] for dielectric materials interacting across air and liquid according to validated methods [54]. Please see Appendix 2 for more detail related to the calculation of Hamaker constants relevant to alkali chlorides in boiler deposits.
Calculations of Hamaker constants between interacting materials follow the protocol of a layer model of material 1 interacting with material 2 through a medium (material 3) and is thus for a three layer model given the terminology A132. For the systems discussed in this report material 1 relates to the boiler tube material, the medium 3 is liquid KCl ; material 2 is air (tables 5 and 6) or alternatively a ceramic coating, or KCl
Air (table 5,6) KCl (table 5,6,7)
Figure 6. Layer model for determination of Hamaker constants A132.
Figur 6. Skikt-modell för bestämning av Hamaker-konstanter A132
26 VÄRMEFORSK
Ninham and Mahanty [55] published the following relationship for the Hamaker constant in terms of a damped oscillator model: