You are on page 1of 11

Journal of the European Ceramic Society xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Journal of the European Ceramic Society
journal homepage:

Original Article

Exposure of refractory materials during high-temperature gasification of a
woody biomass and peat mixture

Markus Carlborgb, Fredrik Weilanda, , Charlie Mac, Rainer Backmanb, Ingvar Landälvc,
Henrik Wiinikkaa,c
RISE Energy Technology Center, S-941 28 Piteå, Sweden
Umeå University, S-901 87 Umeå, Sweden
Luleå University of Technology, S-971 87 Luleå, Sweden


Keywords: Finding resilient refractory materials for slagging gasification systems have the potential to reduce costs and
Gasification improve the overall plant availability by extending the service life. In this study, different refractory materials
Oxygen blown were evaluated under slagging gasification conditions. Refractory probes were continuously exposed for up to
Biomass 27 h in an atmospheric, oxygen blown, entrained flow gasifier fired with a mixture of bark and peat powder. Slag
Entrained flow
infiltration depth and microstructure were studied using SEM EDS. Crystalline phases were identified with
powder XRD. Increased levels of Al, originating from refractory materials, were seen in all slags. The fused cast
materials were least affected, even though dissolution and slag penetration could still be observed.
Thermodynamic equilibrium calculations were done for mixtures of refractory and slag, from which phase as-
semblages were predicted and viscosities for the liquid parts were estimated.

1. Introduction membrane wall, which during operation is covered by a layer of solid
slag over which the liquid slag will flow. This type of wall has the
Biomass gasification can become a part of future energy systems for advantage that it is extremely durable. Almost no corrosion will occur
the production of sustainable transportation fuels, chemicals and because the membrane wall only comes in contact with solidified slag.
power. Among gasification technologies, the entrained flow technology Drawbacks are, however, high investment cost and higher heat loss
currently under development for biomass produces the highest quality (2–4%) compared to refractory walls (1%) [1] which significant re-
syngas, i.e. tar free syngas mainly composed of CO and H2, [1–3]. duces the gasification efficiency.
Furthermore, most industrial coal gasification plants developed after Despite this, slagging gasification systems employing refractory
1950 are of the entrained flow type [1]. Entrained flow gasifiers are walls report refractory life-times of only 6–18 months [1] and extensive
generally operated in slagging mode, meaning that the operating tem- research has been performed to address material issues in slagging coal
perature is above the ash melting point of the feedstock. At this tem- gasifiers. The mechanisms for refractory degradation are related to slag-
perature, tars are destructed and fuel conversion is almost complete. refractory interactions and include chemical dissolution, mechanical
The high operating temperature comes however with the penalty of erosion, chemical and structural spalling [4–9]. The development of
relatively high oxygen consumption. Nevertheless, different types of refractories for coal gasifiers continues to be active, and indicates that
reactor walls have been developed for coal gasifiers to protect the re- the development of refractories for entrained flow gasification of woody
actor shell from the harsh conditions of the reaction zone. The re- biomass must be considered as part of the overall development process.
fractory wall is the simplest, most efficient and lowest-cost design [1]. This is heightened by the fact that woody biomass is generally enriched
Here, a hot face refractory material, which can withstand the tem- in elements such as Ca, K and Mg whereas coal typically has higher
perature and chemical conditions inside the gasifier, is installed to- contents of Al-, Fe-, Si- and Ti-bearing minerals [10]. Since the ash-
gether with one or more insulating layers (back-up layers) inside the forming matter in biomass and coal differ considerably, which thereby
reactor. High quality chromium oxide and/or zirconium oxide based also changes the melting and wetting characteristics of the slags [11],
refractories are employed in coal gasifiers because of their chemical refractory materials developed for coal slags are not necessarily re-
resistance to the coal ash. Another type of wall is the water-cooled sistant to the likely more alkaline woody biomass slags (e.g. [12] and

Corresponding author.
E-mail address: (F. Weiland).
Received 28 June 2017; Received in revised form 4 September 2017; Accepted 11 September 2017
0955-2219/ © 2017 Elsevier Ltd. All rights reserved.

Please cite this article as: Carlborg, M., Journal of the European Ceramic Society (2017),

apart outside of the fuel exit. of the rectangular tube was to reduce the conductive cooling effect from bron 160H (90 mm thick). The gasifier achieve a certain gasification temperature. N2. pilot gasifiers must therefore be operated with relatively higher oxygen feed. and temperatures and syngas composition are found in Tables 2 and 3.25 mm directly after blending. Each sample was fitted with boiler cement into a the central part of the gasifier. Thus. The four O2-inlets are directed so that the Refractory samples were cut as cuboids with dimensions attack angle was 45° towards the central axis. T13 in Fig. In order to 2. The resulting process height of approximately 3. ranging from cheap castables to ex.1. 3 is a photo taken from the top of approximately 40 cm in height according to Fig. The resulting syngas composition was greater variation than surrounding refractory thermocouples. we present the results from the eva. it was allowed to cool down in a nitrogen purged sample holder 1200–1250 °C [16]. C2H4 and material. and selected based on the project group’s powder was fed to the gasifier using constant mass flow of transpor- previous experiences from gasification of black liquor [14] and stem tation air corresponding to 220 ± 10 NL/min (average ± standard wood biomass [13].9 m. O2. with the tip approximately 20 mm from the hot face wall. Fuel feeding rate was 25 ± 3 kg/h. The second graphite sample was heavily Glommersträsk. 1. Exposure of ceramics speed of fuel dosing screws. Refueling of the hopper was repeated every 12 h. Sweden. the probes holding the ceramic for woody biomass gasifiers. the gasifier was preheated over night after they were exposed to gasification of a woody biomass and peat with a ∼100 kW oil burner firing conventional diesel fuel.5 mm inlets concentrically positioned and evenly distributed 90° specifications can be found in Table 4.. CO. and showed sifier as indicated in Fig. to slagging gasification during 6 h. on evaluating the degradation of eight different refractory materials Prior to campaign startup. This fuel affected and therefore removed in conjunction with refueling already mixture was chosen based on a previous study that showed that bark after 7 h of exposure. The purpose hot-face being exposed to the gasification environment was Gouda Vi. The individual fuel compositions are and 45 min. representative of the average conditions inside the gasification zone. This created a jet flame in 13 × 13 × 110 mm. During these Additional knowledge is therefore needed towards understanding time periods. CH4. Oxygen Eight different materials. The gasifier has an inner diameter of 50 cm and a 0. with peat from Norrheden. separated neously during the experiment.M. where it mass. The purpose of these exposures was to heating was accomplished by combusting the pulverized fuel mixture identify critical refractory properties (e.47 ± 0. 2. 1. In this work. tion. shown in Table 1 together with the calculated composition of the fuel tion of refractories caused by interaction from ash derived from bio. Average temperatures measured by these thermocouples are also shown Syngas was continuously sampled from the bottom part of the ga. thermocouples were all of type-K. and maintain a temperature of approximately 1100–1140 °C on the refractory thermocouples that were positioned closest to the sample 2. The refractory thickness) that were mounted at the probe tip. λ. mixture. re- therefore only a brief overview is given herein. Once a graphite sample was removed from the hot fuel alone would not form a flowing slag at typical wall temperatures of reactor. Fuel and experimental conditions gasifier for 26 h. whereas the O2 feeding luation including some conclusions and suggestions on refractory ma. There is a lack of studies concerning the degrada. Probe temperatures were generally lower. see Fig. This included approximately 2. Estimations indicated a flowing slag could be in order to avoid further combustion in the surrounding air. whereas the second round of sam- ples. in Table 2. C2H2. Pilot gasifiers generally have higher propor- tional heat losses compared to industrial full-scale gasifiers. The top formed under the mixing proportion of 70 wt% bark and 30 wt% peat. Eight water cooled sample probes were used simulta- thermocouples at eight different levels along the reactor. Fuel entered the gasifier together with transport air through an Ø 50 mm central exit of the burner.g. deviation). The fuel feeding rate was controlled by the rotational 2. This experimental campaign included 42 h of gasifier opera- reactor outlet [13].3. The pulverized fuel mixture was collected in big-bags. Temperature monitoring was performed by the probe itself. PoraPlot U columns followed by TCD (thermal conductivity detector) This investigation included two sample pieces of each refractory detectors for detection of H2. Fuel pensive fused cast materials. Insulating refractory lining protect the rectangular Al2O3 tube (20 × 20 mm outer dimensions and 3 mm wall outer steel shell from the hot gasification environment. see Fig. 1. compositions and micro. Additional mixture in a pilot gasifier. Carlborg et al. Refractory perature measurement at the rear end of the cuboid sample (Fig. until the refractory temperature in the gasifier reached close to 1200 °C. The first sample piece of each refractory material was exposed C2H6.02 during the gasification experiment. structures) that should be pursued in the development of a refractory Once this temperature was reached. CO2. The locations of the probes were chosen as approximately (T9–T16).5 h of paused operation during hopper refueling and approximately 1 h in combustion mode The feedstock was prepared by mixing a bark fuel from during heat-up after refueling. piece of the second HB-sample broke and fell down into the boiler part The fuel mixture was milled in a hammer mill with sieve size of of the plant after being exposed under gasification conditions for 11 h 1. Material Ø 3. Eight thermocouples the gasifier showing the probes installed inside the reactor during ga- were positioned in the gas phase (T1–T8) and eight in the refractory sification. It was previously described in [15]. that are commercially available and used (O2) was controlled by a mass flow controller and injected through four in different refractory applications. Pulverized fuel was spectively. This was monitored by a Micro-GC (Varian 490 GC) with molecular sieve 5A and most probably an effect of the water cooling of the probes.e. The present study focused stopped. Journal of the European Ceramic Society xxx (xxxx) xxx–xxx references therein). in this work was operated at an oxygen stoichiometric ratio. 1. All other samples were just removed from the gasifier and 2 . of periments. the gasification process was paused by introducing a small biomass slag-refractory interactions and to develop resilient refractory purge flow of N2 through the gasifier while fuel and O2 feeding was materials for slags with origins from biomass. Fuel powder was mullite-based refractory indicated detrimental interactions with woody pneumatically transported from the big-bags to the receiving fuel biomass ash that led to fluxing of the refractory and blockage of the hopper. 2). O2 feeding rate in order to switch the operation to gasification. the gasifier used An atmospheric entrained flow pilot gasifier was used for the ex. while previous experiences from a pilot-scale reactor lined with was stored awaiting the gasification experiments. rate (174 ± 8 NL/min) was used to control the process temperature terials for slagging woody biomass gasifiers. were installed in the 2. samples were installed in the gasifier directly followed by reducing the mercially available materials. except graphite and top piece of HB-sample. were chosen for this study. pneumatically transported from a fuel hopper to the burner mounted on top of the gasifier. Fig. Gas phase thermocouples were protected by ceramic encapsulation (Ø Two of the probes were equipped with type-S thermocouples for tem- 8 mm) and were of type-S (T1–T4) and type-K (T5–T8).2. i. Sweden. Experimental probes. The refractories were chosen from com.

WOLLA. Compared to detector for energy dispersive x-ray spectroscopy (EDS). C3 Pa. 2. Mel_A. Cord. The cuts were made composition. Neph. A cross section taken 1 cm from the outer edge. Morphology and elemental composition of the refractory cross sections was investigated 3. NCA2. cPyrA. Schematic overview and picture of the gasifier with probes positions clearly marked. ZrOc. centrations while Ca. NASh. and Mg in the case of spinel probes indicates refractory dissolution. The slag on the graphite probe had higher Si concentration talline compounds. The pure phase and solution databases selected were FactPS and FToxid (SLAGA. The bulk and identifiable crystalline compositions of each re- probes length was prepared for all probes except one that had bent fractory were studied with calculated phase assemblages with the slag which was prepared parallel to its length instead. C3A1. using Cu K-α radia- possible contamination in considerable amounts from the graphite tion and a Ni-filter on the detector side. CAF6. CAF1. Since the only AXS d8-advance equipped with a våntec detector. probe is carbon.1 [17] and Chemsheet [18] were used to perform Anorthite was the most common new phase and was found in all 3 . C3AF. while the matrix composition of each refractory was stu- with a diamond blade lubricated with mineral oil. solution or hydration. Thermochemical equilibrium and viscosity calculations on other probes. CAFS. MeO_A. The slag composition on all Samples from the affected area and slag on the probes was pulverized probes except graphite had only small variations in composition be- and investigated with powder X-ray diffraction (XRD) to identify crys- tween them. C2AF. It was done with ionized argon accelerated at 8 kV for 3.5. ZrOt. KA_H. thermochemical equilibrium and viscosity calculation in order to aid sphere.4. The enrichment of Al in the slag 2. and Fe was lower. perpendicular to the M2 Pa).1. oPyr. For the fused cast material it was necessary to study some finer details so the SiC paper polishing was complemented by ion milling. aC2SA. pPyrA. Results and discussions 8 h and then at 2 kV for 6 h at a beam angle of 4°. Carn. interpretation of the experimental results. [19] with estimations of the slag melt viscosity. M3 Pa. Elemental composition and morphology SEM EDS in a Zeiss EVO LS15 scanning electron microscope (SEM) with LaB6 electron source and equipped with an Oxford Instruments xmax-80 Slag on top of the probes was analyzed with SEM-EDS. LcPy. The samples were died with a step-wise calculation method introduced by Reinmöller polished with SiC paper without lubricants to avoid the risk of dis- et al. KASH. FactSage 7. Analyses of exposed ceramics CAF3. Carlborg et al. CMPc. Journal of the European Ceramic Society xxx (xxxx) xxx–xxx Fig. Imaging was the ash composition the average slag had increased Al and Si con- done with back scattered electrons (BSE) for atomic number contrast. 1. AlSp. C3Pb. CAF2. CORU. The XRD analyses was done in 2θ mode on a Bruker and lower Al concentration than slag on the other probes. allowed to cool down to room temperature in the surrounding atmo. OlivA.M. this composition is viewed as closest to what is formed solely from the fuel ash in the reactor.

3 ± 0.2 ± 2.1 Ob wt% d.6 5. 1.5 Moisture wt% 10.2 ± 2. a large crack going from the upside and down into the material is visible on the probes outer edge filled with slag.M.h 51.s.8 ± 2.5 mm depth. 630 ± 126 3350 ± 670 1446 ± 219 Sif mg/kg d. 19. EDS-ana. 295 ± 45 105 ± 16 238 ± 32 Fed mg/kg d. Two types of failure caused by volume Gehlenite. mullite or andalusite in considerable between Al. 40. Corundum was the only original and possibly also damaging the containment vessel [20–23]. c EN 14775:2009/EN15403:2011/SS 187171:1984 mod. materials containing corundum. f EN 14385/ICP-AES.s. 590 ± 89 590 ± 89 590 ± 68 Sd mg/kg d. refractory lining in several ways. Element Unit Bark Peat Mixtureg Ca wt% d.s.6 ± 3. 750 ± 150 13000 ± 2600 4425 ± 787 Pd mg/kg d.2 ± 0. After 27 h the sample had been pressure on materials behind it with compressed insulation materials bent and completely infiltrated by slag.s. observed. e SS-EN 14918/15400 ISO 1928.7 3.01 T8 1045 ± 22 T16 933 ± 16 C2H6 < 0. 10–20 μm wide and up to 1000 μm long.2 6.s. but also a large concentration of concentrations.6 ± 0. Leucite phase left while anorthite.7 ± 5. b EN 14918:2010 annex E/EN 15400:2011 annex E/ASTM-D (by balance).-% K and is re- so it is likely that the formed phase identified as gehlenite does not have cognizable from its brighter shade in BSE images and on its lost por- a strict stoichiometry.7 ± 4. h d.6 ± 1.7 ± 0. i Analysed by ALS Scandinavian AB: Ashing at 550 °C followed by fusion with LiBO2 and dissolution in HNO3 and analyzed according to SS EN ISO 17294-1.01 T7 1087 ± 20 T15 1055 ± 21 C2H4 < 0.1 Probe T2 1077 ± 81 CO 23. mainly composed of corundum Ca. Table 3 Gas temperatures (°C) Refractory temperatures (°C) Resulting dry syngas composition during gasification (average ± standard de- viation). Formation of new phases may cause failure of the osity.5 10.s. Measured process temperatures by the thermocouples (average ± standard deviation). = dried sample. where åkermanite (Ca2Mg[Si2O7]) is another endmember. 4 shows infiltrated slag as a bright network between inside the refractories. 54.s. See Table 5 for identified phases in exposed materials and reference 3.2 32.9 N2 35.5 38.7 (mod.s.5 5. The affected matrix contains about 2 at.3 ± 3.3 ± 0.) with EPA-method 200.6 53.11 20. 27 h was oriented with the left side in the figure towards the reactor lyses on these crystals show approximately the correct proportions center. 685 ± 103 830 ± 125 729 ± 81 Ald mg/kg d. Long.8 ± 0. 6.2 ± 0.2 ± 1.s. High alumina with Ca aluminates (HA ± CA) materials.s.2 ± 9.16 19. T1 1246 ± 27 T9 643 ± 35 T2 1276 ± 19 T10 968 ± 20 Gas species Concentration in dry syngas (mol-%) T3 1268 ± 20 T11 1005 ± 16 T4 1219 ± 20 T12 1088 ± 29 H2 13. 2 (mod. A small area of unaffected matrix was left at about 7.3 T6 1114 ± 20 T14 1095 ± 22 C2H2 < 0. 215 ± 43 11000 ± 2200 3451 ± 661 Znd mg/kg d.) with EPA-method 200./ICP-AES. needle-like crystals can be seen in the slag grains.26 1. denoted by a red Table 2 arrow.3 ± 2. formation of gehlenite was also during operation.0 T5 1170 ± 18 T13 1125 ± 22 CO2 23.s.0 Probe T1 927 ± 43 CH4 0. covering almost the entire material. g Calculated based on the proportions of separate fuels. 66 ± 20 383 ± 117 161 ± 38 Mgd mg/kg d. The latter may exert phases corundum and diaoyudaoite. Gehlenite is an endmember of the solid soultions in the me- lilite group.s. and spinel could be found in addition to the original expansion are spalling and expansion of the lining.01 4 .s. 5.s.3 11.1.5 ± 6.6 51.8 (mod.9i 95 ± 19 33 ± 6 Mnd mg/kg d. It is possible that these formed upon cooling of the probes and not but also some calcium aluminates.s.42 Nai mg/kg d.3 Lower heating valuee MJ/kg d.4 Na wt% d. Carlborg et al.s. 0. gehlenite and spinel had been formed.).0 Ha wt% d. Journal of the European Ceramic Society xxx (xxxx) xxx–xxx Table 1 Major elemental composition of the fuel mixture used in the gasification experiments. way through.4 ± 0.6 ± 0.0 ± 0.2 Ashc wt% d. 245 ± 13 2500 ± 125 922 ± 39 Cld mg/kg d.1. The was only found in the spinel samples but does not seem to have formed BSE image in Fig. 2100 ± 630 585 ± 88 1646 ± 442 Cad mg/kg d.s.03 2. After 6 h of exposure the sample showed infiltration almost all the milar and many possible constituent elements are available in the melt. 109 ± 17 180 ± 27 130 ± 14 Kd mg/kg d. Si and K for being leucite.) and SS EN ISO 11885 (mod.s.s. The sample exposed for on these samples.6 ± 4.9 ± 3. The XRD patterns produced from these phases are very si. d NMKL 161 1998 mod. In the HA + CA brick.5 ± 9 37 ± 11 49 ± 7 Bark and peat were analyzed by Eurofins Environment Sweden AB according to: a EN 15104:2001/EN 15407:2011. leucite. 4500 ± 675 5150 ± 773 4695 ± 526 Tid mg/kg d.s. Uncertainty estimated by Taylor series method.

slag had penetrated the material to a depth of about the refractory matrix. 6 a BSE image with Fig. and 27 h of exposure the matrix was 40 μm. MgO 35% periclase < 10%) Other 1% High alumina spinel fused casted (HASPf2) Al2O3 53.1. the dense parts of the fused cast spinel was infiltrated to a depth of about 30 μm. CaO 1. In Fig. SiO2 12% chemical and metallurgic industries.9% Other 1% Silicon carbide low cement (SCLC) SiC 60% Low cement castable base on silicon carbide Al2O3 30% with good thermal conductivity and high SiO2 5% abrasion. Carlborg et al.8% Fe2O3 0. Thermocouple (type-S) position is shown in the picture (bottom probe). filtrated the matrix and crystals (likely anorthite) have formed on the surface. corundum and mullite). Water cooled sample probe with mounted refractory sample (top probe). resistant CaO 4. 4 an overview dition to the original phases (andalusite. MgO 44.1% high temperatures. fused cast material.5% against abrasion. dust erosion or impact at SiO2 0. 5 . Andalusite (ADL) 3. indicated by a red arrow).4.5% refractory applications SiO2 0.2% resistance. In Fig.-% K in For 6 and 27 h.6% Magnesia rich fused cast refractory. Graphite Brick (C) Graphite – 3. Hibonite (HB) Al2O3 90. High alumina spinel fused casted (HASPf1) Slag infiltration is visible as loss of porosity and up to 2 at.1.3. In ad.4% Fe2O3 1. the crystalline phase leucite could be found in addition partial intrusion could be observed through the whole samples. After 6. to the original phases (periclase and spinel). 4 and for 27 h is displayed in Fig. Refractory material Composition (wt-%) Description according to material specification High alumina with Ca aluminates (HA + CA) Al2O3 94% Tabular alumina based castable.6% Hibonite-based castable for high-alkali CaO 8. In Fig. respectively but After exposure.05% Andalusite (ADL) Al2O3 62% Andalusite based. Journal of the European Ceramic Society xxx (xxxx) xxx–xxx Table 4 Material specifications for the tested samples.4% TiO2 1.7 and 3 mm. 2. 4. The matrix had a similar appearance further into the material 3. resistant to slags in glass. severely affected to depths of about 0. no slag intrusion is visible on this scale.1% Isopressing zirconia with mullite (ZR + ML) Al2O3 66% Acidic refractory with low thermal expansion ZrO2 20% coefficient. An overview of the material exposed In addition to infiltration via large pores (displayed in Fig. Slag in- new phases anorthite and leucite could be detected.2.1% High alumina spinel fused casted (HASPf1) Al2O3 64% Void free fused cast refractory (spinel > 90%. the image is displayed. oxidation and thermal shock Fe2O3 0. trusion via pores was not as extensive in this material as in the other terface between slag and refractory is displayed where slag has in. strong castable with high SiO2 33% shock resistance.1. 5 the in. High alumina spinel fused casted (HASPf2) but with more porosity preserved. slag was found in some larger cavities connected to the surface.M. Fe2O3 0.

carried out with the Viscosity module in FactSage 7 and are shown in Fig. The fluidity of the melt and the had been exposed for 27 h. This will cause a material expansion followed by stress and possibly crack formation. the lack of anorthite formation a Fuel ash composition 1. The slag has a darker shade than the refractory parts in these images because the refractory has a higher average elemental composition. diaoyudaoite.4 mm after 27 h.1 16. SCLC and ZR + ML matrices produce very viscous melts Slag penetration was visible as filled voids between the zircon grains with increasing refractory matrix. which was identified slag. due to the lack of components in this probe that could be dissolved by the slag. to a much lesser extent. and Si quantified on a carbon and oxygen dalusite.1 4. phase formation kinetics and transport limitations. 10.2. and about 3. Not only is the density lowered. in the HA + CA-brick. mullite. a TEC of 100 g of the original slag composition and 100 g formed after exposure.-%. in particular Slag infiltration in what seems to be MgO positions were observed to melilite and anorthite may cause degradation. periclase the agreement with the anorthite phase found from the HA + CA. the equilibrium was then calculated again.5 4. 18. 15 calculations). The HA + CA matrix also forms new phases could be detected with XRD. The intruded slag is visible as bright areas between grains in for the HA + CA and HASPf2 matrices in Fig. Silicon carbide low cement (SCLC) Given that the matrix of refractories often interact more extensively After 6 h of exposure the slag had penetrated about 0. A solid solution of melilite was also predicted to form observed how some spinel grains had been completely surrounded by from slag interactions with CaAl2O4 and grossite. 4 and unaffected areas below are darker.8 39. Leucite was also not Slag on graphite probec 2.1. Photo taken from the top of the gasifier showing the refractory samples installed was located at approximately the same level as the exposed material in the reactor. the step-wise calculation method introduced by Reinmöller et al. A temperature of 1220 °C was used in all calculations based on the shielded thermocouple TC-4 that Fig. This is in free basis is shown. displayed in Fig. Hibonite (HB) fractory. slag on of each refractory material and the slag. Dislodged grains could be slag to become more fluid with increasing matrix share in equilibrium seen at the edges of the material. After 12 h the slag had penetrated about 7 mm into the 3. An overview of the exposed a very fluid melt. After 27 h it could be SCLC refractories. Leucite was found after upon their formation (Table 6). but 6 h of exposure and after 27 h. an- elemental maps for Mg. the solid solution augite was detected. No probe did not last as long as the others. and 1. Grains at the edge of the material was also formation of anorthite are possible reasons as to why this refractory observed to being disintegrated into smaller. The phases predicted are demonstrated material. Journal of the European Ceramic Society xxx (xxxx) xxx–xxx surface is displayed in Fig. Phase assemblages were also generated to evaluate the stability of b Obtained from EDS-analyses.7 in the HB-brick indicates that factors besides thermodynamics have an Slag averageb 1.9 9.3 3.1. Spinel could be detected Estimations of the molten slag viscosities for each calculation were in addition to the original phases (hibonite and corundum). Anorthite had been Initially.9 predicted for slag interactions with MgO nor spinel.5 15. 7. No new crystalline phases could be detected with XRD in crements until a final molten slag to refractory matrix ratio of at least this sample.6 24.1. The formation of these phases.M. the HB matrix interacts with the probes exposed 6 h and 27 h.5 mm into the material for would be limited. due to changes in density a depth of about 40 μm after 6 h of exposure. more pronounced for the material that with anorthite.4 4. corundum and hibonite. respectively.9 13. 4 and a detailed image of the material smaller amounts of anorthite and melilite.0 7.5 mm.4 6. These peaks were weakened in the exposed material and no phase could be assigned with certainty. A sharp transition could be observed between in. On the other hand. hibonite and.6.e. phases identified from XRD but differences are expected given the Na Mg Al Si P K Ca Fe heterogeneous make-up of the refractories. respectively.5 3. corundum. c Obtained from ICP-analyses. average of slag on spinel materials presented on an carbon and oxygen predicted are listed in Table 7. in addition to materials is displayed in Fig. Under these conditions.9 influence over the slag refractory interactions. These findings indicate that the binding mullite phase is being dissolved by the slag with loosening of grains on the surface as a result. in equilibrium with mainly hibonite. In the BSE image. Two small peaks of what seems to be mullite could be seen in the XRD pattern for the unexposed material.1 mm into the repeated with the amount of refractory matrix increasing by 20 g in- refractory.9 54.9 8. suggesting that slag penetration and was observed 2. and together with 120 g of re- 3.8 1. These were based on attacked by slag. filtrated and unaffected matrix. large grains did not appear to have been were carried out using the matrix compositions. 8.6 17. Carlborg et al. For example. Note that the graphite sample was removed before this photo was taken. Fig. 3. refractory matrix was carried out. 9. were the main phases predicted. Al.9 6.9 mm into the with the slag and facilitate penetration. Table 5 Phase assemblages were generated based on the bulk compositions Elemental composition of fuel ash and average slag composition on all probes.4 5. 3. A gasification atmosphere corresponding to the measured gas thereof the empty position in the top of the image. material is also added.3 43. probes. the 12 crystalline phases identified in the pristine refractories. spinel.1 11. Anorthite was predicted to be formed as a major crystalline phase from slag interactions with CaAl2O4. slag is the brightest. The resulting molten slag composi- tion was then normalized to 100 g. Instead olivine and sapphirine.0 32. TECs and viscosity estimations matrix.9 6. composition was also fixed. The major (and minor) phases graphite probe.8 7.6 9. [19].0 12. a Calculated from major element composition. grossite. They are mostly in agreement with the free basis in at.8 g/g (i. taken as true slag composition. the slag is pre- dicted to be completely molten within a two-phase melt. 3. This procedure was After 6 h of exposure slag had penetrated about 2. Isopressing zirconia with mullite (ZR ± ML) The ADL. 6 . Thermodynamic equilibrium calculations The slag composition from the graphite probe was assumed to be the true composition of the ash slag.6 Slag on spinel materialb 1. ADL and darkest and spinel intermediate shade of gray. more Zr rich grains.0 2.

The top of the sample images have been oriented upwards in the reactor. they become compatible with spinel and MgO with increasing amounts of refractory. SEM overview images made with back scattered electron detector of material samples made at 20 kV electron acceleration voltage.) Although the HASPf1 and HASPf2 refractories result in very low viscosity melts. This suggests that they will flow and fill voids. Protection of silicon car- bide grains is likely to be acting in a similar way. the reader is referred to the web version of this article. Depending on the dissolved components the slag viscosity. may increase or decrease. which is at 75% relative the others. 3. 4. and all have the same scale.M.3. Potential crystalline phases The results from XRD analysis are summarized in Table 7 together with predicted phases from TEC. Difference in atomic number gives contrast in these images. refractory components dissolve and change the slag composition. the viscosity of intruded slag increases as more refractory components are Fig. Discussion When ash slag comes in contact with the material probes. 3. Carlborg et al. The red arrow indicate a large crack in the HA + CA material and a pore filled with slag in the HASPf2 material. and the slag appears brighter due to its heavier average elemental composition except for ZR + ML where the slag is lighter. where Si is abundant. except HA + CA 27 h. Journal of the European Ceramic Society xxx (xxxx) xxx–xxx Fig. but will not dissolve components of the refractory extensively. and approximate limit for severe slag infiltration in the ADL material. (For interpretation of the references to colour in this figure legend. In the case of si- licon carbide castable and andalusite castable. as the oxygen activity 7 . Red bars denote slag infiltration depth. and therefore the con- tinued rate of infiltration. 5. incorporated and practically comes to a halt. SEM image of interface between slag and ADL refractory. 5000 μm scale bar shown.4.

An alternative scenario that tective oxide is formed [24]. 7. Intruded slag could. slag. When the binder phase is replaced with slag the zircon changes in composition of the matrix and changes in microstructure. a layer of pro. Removal or fluxing of this layer would could bend the material would be if a large portion of new crystalline lead to increased wear of the grain. Interface between slag and fused cast spinel (HASPf2). composed of corundum or hibonite. The large cracks in the upper part. Higher Si content and larger grains should delay the effect because Fig. Periclase grains are embedded in a spinel matrix. the viscosity of the slag does not with a downward bend as effect. and Si on carbon and oxygen free basis. filled with slag. Journal of the European Ceramic Society xxx (xxxx) xxx–xxx Fig. 6. EDS data was collected at 10 kV acceleration voltage. Carlborg et al. The slag is visible as a brighter shade and it has penetrated the refractory to a depth of about 40 μm in what appears to be former MgO positions in the refractory. Al. increase and therefore the infiltration is deeper in these materials. 8. Back scattered electron image and concentration maps for Mg. in the gasifier is high enough to oxidize the carbide. Fig. 8 . This effect should be by matrix dissolution that has gone so far that the bulk material loses its seen for all materials where the matrix is being dissolved by intruded rigidity and bends under gravity. As the binding phase in the zircon brick (ZR + ML) is being dis- cause of the fine microstructure and the intimately mixing of slag and solved the slag viscosity is initially increased but as more original slag is matrix it is hard to isolate intruded slag when performing elemental incorporated in this mixture the viscosity is approaching that of un- analysis in SEM. the slag. Be. SEM image of ADL material. The horizontal field of view is 115 μm. be detected by small altered slag. Some grains are seen in the slag after 6 h of The bending of the corundum castable (HA + CA) could be explained exposure and the effect is more distinct after 27 h. speak for this explanation. Slag and refractory interface with infiltrated matrix. grains becomes mobile.M. however. When the castables are mainly phases is formed on top of the material while the bottom expands less. Surface of the zircon mullite refractory with dislodged and dissociating grains in In the lower part the matrix is unaffected by slag.

Zhang et al. Ca.26 Periclase 3. Grossite 2. 6 are similar Leucite 2.0 pletely embedded in spinel has been dissolved and slag has taken its Mullite 3.M. 9 .65–2. Even though slag infiltrates and to some extent dissolves it.21 diction with the TEC predictions. size and orientation of the slag areas just beneath the Hibonite 3.6–4.57 place in the material.1 served at far lower temperatures [25–27]. After 27 h. Periclase that is not com- Melilite 2.9–3. Ti)(Si. The small grains at the surface shows higher Mineral Density [g/cm3] concentration of Zr than the large grains but also some Si.66 it could be observed how spinel grains were completely surrounded by Spinel 3. while ZrO2 material seems to be the least affected among the tested materials.76 1673 ± 10 °C [25] but in the presence of impurities it has been ob- CaAl2O4 2. This means that even these dense materials are risking to be dis- Zircon 4.74–2. this ZrSiO4 was predicted to be stable against the slag. in addition to the dissolved periclase. Pure zircon dissociates into oxides at Anorthite 2. Quartz 2.11–3. [28] found anorthite after exposing alumina to a model slag rich in Ca. and other elements found in the slag. The formation of anorthite was predicted from TECs and also ob- served in Al-silicates and the corundum material. surround large grains. Journal of the European Ceramic Society xxx (xxxx) xxx–xxx Table 6 more slag is required to reach viscosities where grains start to move and Density of minerals identified in pristine and exposed re. Al. the liquid zone must stretch deeper into the material to completely fractories. Refractory Unexposed material 6h 27 h TEC phase assemblage High alumina with Ca CaAl2O4 Al2O3 (corundum) CaAl2Si2O8 (anorthite) CaAl2Si2O8 (anorthite) aluminates (HA + CA) Al2O3 (corundum) NaAl11O17 (diaoyudaoite) Al2O3 (corundum) Ca. would lead to the formation of ZrSiO4. Al2O3 (corundum) Al2O3 (corundum) Cordierite cristobalite) SiC (different types) SiC SiO2 (quartz) Mullite SiC Tridymite Hibonite (HB)a Al2O3 (corundum) Al2O3 (corundum) Al2O3 (corundum) CaAl2Si2O8 (anorthite) CaAl12O19 (hibonite) CaAl12O19 (hibonite) CaAl12O19 (hibonite) Corundum MgAl2O4 (spinel) Hibonite Ca.55–3. Carlborg et al. Fe.88 The shape.98–4. solid solution) MgAl2O4 (spinel) MgO (periclase) KAlSi2O6 (leucite) Spinel MgAl2O4 (spinel) MgO (periclase) Monoxide MgAl2O4 (spinel) Olivine Silicon carbide low cement Al2O3 (corundum) CaAl2Si2O8 (anorthite) CaAl2Si2O8 (anorthite) CaAl2Si2O8 (anorthite) (SCLC) SiO2 (quartz.6–4. quartz) a The would-be 27 h sample broke and fell out of the gasification chamber after approximately 12 h.5 to the periclase areas within the material.Mg-Aluminate (Spinel) (Leucite) Isopressing Zirconia with Al6Si2O13 (mullite) ZrSiO4 (zircon) ZrSiO4 (zircon) CaAl2Si2O8 (anorthite) mullite (ZR + ML) ZrSiO4 (zircon) ZrO2 (zirconia) ZrO2 (zirconia) Corundum ZrO2 (zirconia) ZrSiO4 (zircon) ZrO2 (zirconia) Mullite (Sapphirine) Graphite Brick (C) C (graphite) SiO2 (Cristobalite.13–3. This observed disintegration is in contra- Andalusite 3.7 assembled from long time exposure in a similar way as the other ma- Zirconia 5.94 Corundum 3.45–2. Al)2O6 Sapphirine (Mg4Al10Si2O23) casted (HASPf2) (augite.83–3. Si.6–6 terials.1 slag. Na)(Mg.Mg-Aluminate NaAl11O17 Ca2Al2SiO7 (gehlenite) Ca2Al2SiO7 (gehlenite) Corundum (diaoyudaoite) CaAl4O7 (grossite) MgAl2O4 (spinel) MgAl2O4 (spinel) CaAl12O19 (hibonite) (Spinel) (Leucite) Andalusite (ADL) Al2SiO5 (andalusite) Al2SiO5 (andalusite) Al2SiO5 (andalusite) CaAl2Si2O8 (anorthite) Al2O3 (corundum) CaAl2Si2O8 (anorthite) CaAl2SiO8 (anorthite) Mullite Al6Si2O13 (mullite) Al2O3 (corundum) Al2O3 (corundum) (Cordierite) (Tridymite) High alumina spinel fused MgO (periclase) KAlSi2O6 (leucite) KAlSi2O6 (leucite) Sapphirine (Mg4Al10Si2O23) casted (HASPf1) MgAl2O4 (spinel) MgO (periclase) MgO (periclase) Spinel MgAl2O4 (spinel) MgAl2O4 (spinel) Monoxide Olivine High alumina spinel fused MgO (periclase) KAlSi2O6 (leucite) (Ca.85 surface of the fused cast spinel-periclase displayed in Fig. Table 7 Identified crystalline phases in samples.

it may be resilient for invaluable assistance before. This is calcite. spalling. Gasification. Wiinikka. Upon heating of this mixture gehlenite was formed at tem. Conclusions Öhman is also thanked for his insightful comments and critique to improve this manuscript. Calle Yllipää. Al-Si spinel. GPP. and anorthite from heating kaolinite and Si content showed less intrusion than those with low Si content. formation from corundum in contact with Ca. Öhrman. 9. Even though one Esbjörn Pettersson and Mattias Lundgren are also highly acknowledged refractory might not be thermodynamically stable. during and after the experiments. Heidenfors at Fagersta Eldfasta is also gratefully acknowledged for supply and helpful advice regarding materials to be tested. Formation phase detected in the hibonite castable even though TECs predicted of hibonite was not seen in this study even though it was predicted by mainly anorthite and Ca-Mg-aluminate. Marklund. be ranked and assessed when choosing a material. Bo enough to have an acceptable time of service.M. Viscosity estimation of melts penetrating refractories. [29] studied formation also spinel was dissolved from the fused cast spinel. Marcus 4. Weiland. with ∼8% Si. Schaafhausen et al. and Fe at a temperature of 1600 °C. which is highly acknowledged by the authors of this dissolution and dislodging of grains) taking place in refractories should work. the density of the products are lower than the ones in the ori. 10 . mullite castable. Carlborg et al. Anorthite was formed in the corundum castable.28. viscosity. Ptáček et al. system. M. Fig. Yngve Ögren. R. H. Journal of the European Ceramic Society xxx (xxxx) xxx–xxx Fig. Henry Hedman. These experiments were however done with temperatures above 1500 °C as compared to about 1220 °C in this study. Other researchers have observed hibonite spinel. and < 2% Al) at 950 °C and 800 °C. means that Al is being dissolved from all materials. Higman. 10. M. Castables with high of gehlenite. Prof. All tested materials showed signs of wear after 6 and 27 h exposure but fused cast spinel seemed least affected in terms of slag intrusion and References formation of new phases. The slag on refractory probes all had higher Al concentrations than ash slag collected on a graphite probe. 2008.31. which [1] C. and SiC-corundum casTable Spinel was the only new ginal refractory which means that the volume will increase.32] and also from letting natural dolomite decompose and react with cor. 2017. with only minor levels of TECs for some materials. slag [19. van der Burgt. Gebart. This work has been founded by the Swedish Energy Agency through The severity of the mentioned destructive effects (swelling. These grains also dissociated which was in contra- of these phases. Hedman. besides that more mass in form of CaO is added to the diction with the TECs. Jonas Wennebro. O. attributed to the altered slag composition followed by changes in peratures above 950 °C and anorthite at 1256 °C. Acknowledgments undum [33]. Mainly periclase but [2] F. During formation material surface. The zircon brick showed signs of failure by dissolution of the [30] found gehlenite after exposing mullite to wood ash (mainly Ca and binding mullite phase which led to removal of zircon grains from the K. Bio4Gasification. Phase distribution for slag/refractory matrix interaction (left) HA + CA and (right) HASPf2.

Neuroth. Gullberg. Schaafhausen. H. Ceram. F. M. Kim. Sarpoolaky. in: Y. Trans. T. [12] V. Technol. Pelton. Ceram. J. (5) (2015) 1613–1620. Moya. Ceram. Oxid. Energy Fuels 30 (12) (2016) aluminate spinel-rich refractories by sulphur-containing slag. [8] R. J. Backman. Jung. 149 (2016) 218–230. Eur. C. Bale. A. [4] K. Phase Diagr. Mater. R. polycrystalline corundum and calcium hexaluminate by calcium silicate slags. Mehofer. Hedman. Soc. Reinmöller. Parr. Failure mechanisms in high chrome oxide gasifier re- ZrO2-addition on the interaction of alumina castable with molten coal and gasifier fractories. Caballero. [24] E. M. P. Gehre. S. Fuel Process. A critical review of mineral matter related issues [27] P. [11] G. Alumina dissolution into silicate slag. Carlborg et al. Reinmöller. Eur. Corrosion [15] J. A. Advantages and disadvantages of composition tion reactions of silicon and silicon-carbide. phase crystallization from calcined kaolin. Nakano. Black Liquor Gasification. Koukkari. Penttilä. Bélisle. Krishnamoorthy. Chartrand. Kang. Mechanical Properties and Processes – Computer Simulation and Modelling. Ash melting behavior and slag in. M. flow biomass gasifier. Kwong. E. Wennebro. E. Rådberg. Pena. [26] T. Soot concentrations in an atmospheric entrained flow Ceram. Zhang. de Aza. 54 (35-53) (2016) Fuels 27 (2) (2013) 932–941. E. A. Analysis and prediction of slag- 503–513. Nasdala. X.-S. Wiinikka. The kinetics of Al–Si spinel Energy Fuels 28 (11) (2014) 6941–6952. Metall. Components and Refractory Lining. M. [16] C. Gorog. Lee. L. J. Öhrman. 4595–4604. M. Meyer. J. Ceram. Molinder. M. [20] J. [29] P.). H. 29 (8) (2009) 1347–1360. Gheribi. Efficiency and Renewable Energy. Blanpain. J. Wiinikka. H. 17 (7) (1997) 935–941. H. Int. Libowitzky. Kwong. J. Thomas. M. Reactive coating of dolomite on alumina substrates. J. Soc. 8 (2015) 10430–10463. J. A-Phys. Energy 2010–2016. H. H. T. K. Ma. Bennett. M. Hedman. R. Ögren. Walch. Bennett. V. Hack. K. [23] A. R. Umeå University. W. L. Calphad: Comput. Xu. News 9 (5) (2004) 20–25. 330–350. 42A (4) (2011) slag. Eur. Petersen. J. and biomass ash under simulated gasification conditions. Technol. J. Petkov. Petty. Aneziris. H. Soc. E. Jones. Jansson. C. 35 10543–10554. On the filtration into alumina refractory simulating co-gasification of coal and biomass. 97 (1-2) (2009) Fuel 139 (2015) 457–465. J. J.-G. Robelin. Decterov. K. 2007. United States. Fuel 158 (2015) [25] A. R. [28] S. Vassilev. E. P. Sci. Changes in microstructure of a high chromia refractory du to in. V. Sci. Soc. P. Oak Ridge. K. Hack. pp. Ceram. Petersen. Sainz. refractories in slagging gasifiers. Aneziris. Bennett. Ash formation in pilot-scale pressurized entrained-flow [32] P. Wirth. H. E. A. Klinger. Int. during gasification of coal in fixed. C. 34 (3) (2014) 849–876. 2016. Vassileva. Y. and properties of biomass in comparison with coal: An overview. J. 4 (3) (1972) 181–201. Petrol. M. High-temperature oxidation. P. Gebart. [18] K. and entrained flow gasifiers. H. Wennebro. 183 (11) (2010) [14] M. Bengtsson. Y. Bréchet (Ed. Mater. Oh.-C. Medvedovski. Ceram. M. Carlborg. H. Eriksson.-A. A. Eur. Öhman. teraction with infiltrating coal slag in a slagging gasifier environment. Sepman. M. Hydrogen Energy 36 (7) (2011) cesses: a critical review. [9] J. 138–144. K. Leary. of alumina and mullite hot gas filter candles in gasification environment. Weiland. Hager. Thieme. Vazquez. Weiland.-E. Bennett. Öhman. gasifier with variations in fuel and burner configuration studied using diode-laser [31] B. S. F. F. R. 323–330. Refractory liner materials used in slagging gasifiers. H. Eur. induced corrosion of chromium oxide-free refractory material during fusion of coal [5] H. Ma. The zircon thermal-behavior – effect of impurities. J. Volume 3 Wiley-VCH Verlag GmbH. Soc. Kwong. S. 2006. E. Proc. 25 (3) (2004) 547–552. Wear mechanisms of chromia EUROMAT 99. Klinger. Telle. J. Thermal stability of zircon (ZrSiO4). 4 (6) (2007) [19] M. M. T. de Aza. Melacon. Zhang. J. molten slag on the refractory surface. Weinheim. Pettersson. Refract. Gehre. M. Journal of the European Ceramic Society xxx (xxxx) xxx–xxx Pressurized oxygen blown entrained-flow gasification of wood powder. Rezaie. Coupl. 11 . Spencer. S. Fuel 153 (2015) 510–519. Sci. Vassilev.-S. P. Ceram. 2000. Heidenreich. Eur. Umeå. Klinger. B. FactSage thermochemical software and databases. J. Pena. Mueller. A. Wiinikka. B. J. Wang. Pisupati. Carlsson. Gulbransen. Berek. [33] A. 129–138. C. Met. Šouka. Appl. Opravil. Department of Energy. breakdown of zircon upon "dry" thermal annealing. Moss. J. M. 43 (2) (2015) Appl. R. CHEMSHEET – an efficient worksheet Influence of process parameters on the performance of an oxygen blown entrained tool for thermodynamic process simulation. Chinn. R. [30] S. [7] P. Krabbe. S. [3] F. Energy Fuels 30 (2016) 2174–2186. Kaiser. de Aza. Eng. Fuel 150 (2015) 252–260. Lobert. Sangster. Váczi. Havlica. M. Microstructures. B. Metallurg. Improved Materials for High-Temperature [6] E. S. P. Y. Office of Energy sifier environment. Peña. Ceram. 33 (15-16) (2013) 3301–3312.and [21] J. Schreiner.. Simonsson. Influence of TiO2. Corrosion mechanism of extinction measurements. Corrosion of magnesium gasification of bark and a bark/peat mixture. Interactions of refractory Degradation mechanisms and use of refractory linings in copper production pro- materials with molten gasifier slags. Hemrick. 28 (11) (2008) 2199–2211. Mineral. reduction. Pandelaers. Soc. Weiland. Keiser. Reinmöller. Am. Solid State Chem. M. S.-B. H. Slag formation during oxygen-blown entrained-flow gasification of stem wood. Thermochem. Lin. S. Soc. Kwong. fluidized. 888–904. Corrosion resistant refractory ceramics for slagging ga. Energies 19 (1) (1984) 135–142. Black Liquor Gasification – Experimental Stability Studies of Smelt 2565–2569. S. Morphological and interfacial characterization of [22] J. Mater. Yazhenskikh. J. P. Technol. 83 (4) (2000) 897–903. Van Ende. Wiinikka. J. Ptáček. Lotfian.M. 34 (8) (2008) 2107–2116. Nosková. M. R. Sridhar. L. Meyer. Malfliet. Bladh. M. J. [13] P. L. C. Fuel Chem. O. [17] C. Int. G. Scheunis. and volatiliza- [10] S. A. Pettersson. Ceram.