Inhibiting the Carbon Deposition from the Reverse Boudouard Reaction In

Home , Boudouard reaction, Filamentous carbon

TECHNOLOGY TRENDS

Inhibiting the Carbon Deposition from the Reverse Boudouard

Reaction in Refractories Submitted to CO–H2 Atmosphere

J. Poirier, J. Kadok, N. Bost, A. Coulon, M. R. Ammar, S. Brassamin, C. Genevois

The catalytic decomposition of carbon monoxide via the reverse Boudouard reaction in the presence of 2 a mixture of CO–H2 gas produces a solid sp carbon deposit, which has deleterious effects on refractory materials. The catalytic effects of iron oxides and the structural evolution of carbon during the reverse Boudouard reaction in CO–H2 atmosphere is investigated at a nano and micro-scale. Sulphur compounds inhibit the carbon monoxide dissociation on contact with iron and iron oxides. The samples were characterised by X-ray diffraction, Raman spectroscopy, scanning and transmission electron microscopy. A mechanism that governs the inhibition of the reaction is proposed and validated at lab scale, in which the formation of a very thin protective FeS layer (0,5–1 nm) is involved. This research is applied to develop ew CO/H2 resistant refractories.

particles are present in the refractories as From reactions (1) and (2), the water-gas 1 Introduction secondary mineral phases in raw materials shift reaction (WGSR) describes the reaction The mechanism of carbon monoxide de- and as impurities due to crushing and mix- of carbon monoxide and water vapour to composition, called the reversed Boudouard ing while blending the refractory formula- form carbon dioxide and hydrogen reaction: tion. CO + H2O = CO2 + H2

2CO(g) → CO2(g) + C(s) results in the depos The CO resistance is usually improved by DH° = –41,166 kJ/mol 2 ition of sp solid carbon. This C deposition, the selection of raw materials with a low K3 = P CO2 · P H2 /P CO · P H2O (3) inside the porosity of the refractory, causes content of iron particles and by increasing premature degradation of linings, which the refractory sintering temperature (Fe ions In direct reduction processes, the decom operate under reducing atmosphere [1]. The are adsorbed in the lattice or bonded in the position of methane into carbon also occurs destruction of refractory materials in CO at- glassy phase). However, these solutions are CH4 = C + 2 H2 mosphere is well explained in the literature not always very effective for industrial appli- DH° = 92,676 kJ/mol. 2 [2]. cations (such as low-CO blast furnace pro- K4 = P H2 · aC/P CH4 (4) The refractory damage process includes four cess, biomass gasification reactors, steam steps: methane reformers, direct reduction shafts, • diffusion of CO gas diffusion into the por reactors in the petrochemical industry, etc.) J. Poirier, J. Kadok, N. Bost, A. Coulon, osity of the refractory material; where refractories are subjected to CO and M. R. Ammar, S. Brassamin, C. Genevois

• reduction of iron oxides into iron then H2 extremely reducing atmospheres. CNRS, CEMHTI UPR 3079

carburation into Fe3C, which then acts as According to the literature, two reactions University of Orléans a catalyst for C-decomposition according explain the carbon deposition phenom- Orléans, France

Fe3C → 3 Fe + C; ena in a temperature range between 400– • nucleation and growth of the carbon dep- 1050 °C: Corresponding author: J. Poirier osition, which generate thermomechanic • the reverse Boudouard reaction E-mail: [email protected]

al stress inside the refractory; 2 CO = C + CO2 • formation of cracks, damage then de- DH° = –172,4 kJ/mol Keywords: Boudouard reaction,

struction of the refractory. K1 = P CO2 · aC /P CO2 (1) carbon deposition, refractory damage, This reaction occurs at temperature rang- cementite, sulphur ing between 400–900 °C with a maximum • the reverse water gas reaction intensity around 600 °C and is highly fa- CO + H2 = C + H2O The paper was awarded with voured by the presence of catalytic particles DH° = –131,3 kJ/mol the 1st Prize for Posters at ICR 2018 such as iron and iron oxides. FexOy and Fe K2 = P H2O · aC /P CO · PH2 (2) refractories WORLDFORUM 11 (2019) [2] 83 TECHNOLOGY TRENDS

rate of C-deposition occurs from ~530 °C gas, was similar to the gas mixture used in

for 0,8 % H2 to 630 °C for 19,9 % H2. In- low-CO blast furnace industrial processes. creasing the hydrogen content increases the Solid sulphur and sulphur dioxide are used amount of carbon formed. as inhibitors of the Boudouard reaction This paper reports studies of the carbon (Tab. 1). formation by the Boudouard reaction using Laboratory experiments were also carried

FexOy catalysts, and a CO + H2 atmosphere, out with Fe2O3 powders mixed with solid

at 600 °C. The reaction of the carbon for- sulphur and with SO2(g). SO2(g) was mixed

mation was investigated, in these extreme continuously with the reducing CO/H2 gas.

conditions, using thermogravimetric meas- Measurements of Fe2O3 mass variation urements and in situ Raman spectroscopy. were performed using a thermobalance After cooling, the samples were character- from NETZSCH STA 409C/CD coupled

ised using various methods: X-Ray Diffrac- with a QMS 403/5 Skimmer. Fe2O3 sam- tion (XRD) and Scanning and Transmission ples of approximately 30 mg were placed Electron Microscopy (SEM and TEM). in an alumina crucible and were heated Fig. 1 Carbon deposition on a Fe O 2 3 The aims of this research are: to 600 °C at a rate of 10 C · min–1 under sample after exposure to a gas mixture • to understand the effect of iron oxides on 100 ml · min–1 of an argon inert gas prior consisting of 71,0 % CO, 3,0 % CO , 2 the carbon formation and to characterise to being exposed to a reducing CO/H gas 11,0 % H and 15,0 % N , for 5 h, 2 2 2 -1 at 600 °C the structural organisation of carbon ob- (100 ml · min ). tained by the reverse Boudouard reaction The crystalline phases were analysed with

In reactions (1–4), K1, K2, K3, K4 are the in CO and H2 atmosphere; powder X-ray diffraction using a Bruker 2 equilibrium constants of reactions, P CO, P CO2, • to investigate the inhibition of the sp car- D8 A25 diffractometer (Bragg-Brentano

P H2, P H2O, P CH4 the partial pressures of CO, bon formation in refractories from CO dis- geometry θ-θ) equipped with a LynxEye

CO2, H2, H2O, CH4 and aC the carbon activity. proportionation catalysed by iron oxides XE detector and using the Cu Kα radiation

The effect of H2 in CO increases significantly under the most favourable conditions of (λ = 1,5418 Å). Raman spectroscopy was the reaction of carbon deposition. For ex- the reaction, i.e., in the presence of H2 at performed using a T64000 Jobin-Yvon ample, Walker, et al. [3], report a deposition 600 °C; multichannel spectrometer equipped with of 10 g of solid carbon after ≈350 min over • to describe in detail the mechanisms of an Ar–Kr laser source, aBX41 Olympus

0,10 g of carbonyl iron in 9 %-H2–91 % CO carbon inhibition. The effect of sulphur on microscope and a liquid nitrogen-cooled gas mixture at 602 °C. the Boudouard reaction, catalysed by iron CCD detector.

Fig. 1 illustrates this reaction. Few grams oxides in presence of H2 gas, was investi- The spectra were collected in a backscatter- of Fe2O3 were submitted to 71 % CO, 3 % gated under laboratory conditions; ing geometry under a microscope (×20 long

CO2, 11 % H2, 15 % N2 gas mixture during • to find an efficient solution to inhibit the working distance objective) using the 5 h at 600 °C under atmospheric pressure. carbon deposit and to limit the industrial 514,5 nm wavelength (2,41 eV; 3 mW of A large amount of carbon formation was degradations of refractories. laser power), a 600 groves · mm–1 grating observed similar to a “popcorn” formation. giving a spectral resolution of 3 cm–1 in However, this phenomenon is not described 2 Materials and methods the 1000–2000 cm–1 wavenumber range. in CO + H2 atmosphere and the mechanism Various iron oxide particles were used as Heating the sample was achieved using a is not really understood. the catalyst of the Boudouard reaction. The TS1500 Linkam device (Tadworth/GB).

The hydrogen content in the CO has an effect reducing gas mixture (71 % CO, 3 % CO2, The same temperature program was used on the temperature at which the maximum 11 % H2, 15 % N2), referred to as CO/H2 for Raman spectroscopy and for TGA (10 °C · min–1 until 600 °C). After the experiments, the samples were also charac- Tab. 1 Origin and purity of the raw materials terised by scanning electron microscopy Materials Origin Purity Particle Size Specific Surface (Hitachi S4500 FEG) and high resolution 2 [mass-%]) [µm] Area [m /g] transmission electron microscopy (Philips Catalysts Fe Sigma Aldrich >99 2–5 2 CM20 microscope). FeO Sigma Aldrich 99,9 2–5 6 Scanning transmission electron micros- Fe3O4 LKAB >99 2–5 2 a copy high angle annular dark field (STEM- Fe2O3 Sigma Aldrich >99 2–5 8.0 Sigma Aldrich >99 0,0035 a 42 HAADF) images were acquired in the Carajas >98 160–250 10,5 angular range of 50–180 mrad with an Carajas >98 1000–2000 3,7 8 cm camera length and a 0,1 nm probe Reducing gas CO/H Air Liquide >99.9 2 size. Elemental composition line scans

Sulphur dioxide SO2 (g) Air Liquide >99,9 were performed by STEM-EDS using a b Solid sulphur S Alfa Aesar >99,5 ≤10 JEOL EDS system and a 0,13 nm probe a Data as provided by the Sigma-Aldrich Company, b Prepared by grinding in an agate mortar size.

84 refractories WORLDFORUM 11 (2019) [2] TECHNOLOGY TRENDS

3 Effect of iron oxides

on the carbon formation 6

in CO, H2 atmosphere [2]

3.1 Effect of Fe2O3 grain size on ...... - .. '"' the rate of carbon formation . • .. ,,,,,,' Fe2O3 with different grain sizes (35 nm, , <5 µm, 160–250 µm and 1–2 mm/purity >99 %) was used. Initial specific surfaces of the particles are below 10 m2/g, except- ed for the 35 nm size (42 m2/g). Fig. 2 a shows the variation in the sample weight (full lines) based on the thermogravimetric o��-�-�-�-�-��-�-�-�-- measurements (TGA) as well as the vari o 20 2 4 0 ation in the temperature (black dashed line) Tim mm as a function of time. At 600 °C, the sam- Fig. 2 a–b a) Mass variation of Fe2O3 samples versus time depending to the temperature ples were subjected to a CO/H2 gas mixture (in black dotted line); b) mass normalised curve

(composition R: 71 % CO, 3 % CO2, 11 %

H2, 15 % N2) for 5 h. the chemical reduction of the FexOy phase been described in the literature [3, 4]. In a

As soon as the gas was injected (at ap- occurs. After this reduction, a rapid increase mixture consisting of CO+H2 gas, there are proximately 130 min), a weight loss (ap- in the weight of all of the samples was ob- many competing reactions including chem proximately 25 mass-% for the nanometre served, except for the FeO sample, which re- ical reduction of iron oxides, carbon deposit grain size) that was associated with the quires a longer reaction time (≈50 min). The via the reverse Boudouard reaction, and

reduction of the Fe2O3 phase in competition reaction produces a large amount of solid water and carbon formation via the reverse with the production of solid carbon was ob- carbon (Fig. 3 c). water-gas reactions. It is impossible to dis-

served. After this reduction, a rapid increase The FexOy particles are reduced to metallic tinguish these different reactions in thermo- in the weight of the samples was observed iron by the hydrogen. This phenomenon has gravimetric measurements. (Fig. 2 a). This weight gain was primarily due to the production of solid carbon and corresponded to a catalyst/carbon depos ition ratio of approximately 1/100 within the .. 600

asymptotic limit of the Fe2O3 sample. Fig. 2 b 1200 shows the normalised weight curve using an Figure 2 c 500 asymptotic function of the curves shown in � Fig. 2 a. According to the data, the produc- 4 (!) 000 tion of carbon appears to be independent of 800 � the grain size of the catalyst; the size of the ..c B initial grain is not an important parameter .... 300 �(!) ..c 600 for controlling the rate of carbon depos ·- (!) 200 r< ition. However, the smallest grains (35 nm) 4 0 were slightly more reactive and formed more

carbon than the largest grains. The largest Figure 2 b 100 200 � grains exhibited the slowest formation and 1- produced less carbon. .L_�-'-� o � L � ----; - �� ; � 0 0��5 100 L.._1s15 o9 --;2;;;00 2so 30 350 3.2 Effect of the iron valence on n) �T�ml�se (mi f5' 160 --r--�--- the carbon deposition rate \!d) r� 140

Pure iron and iron oxides (FeO, Fe2O3 and

Fe3O4) powders were submitted for 5 h at 600 °C under reducing atmosphere (com-

position R 70 % CO, 10 % H2, 3 % CO2 and 17 % N ). 2 60 120 125 145 1imcs (min) The gas flow was 4,2 l/h. Results are report- 130 135 140 150 ed in Fig. 3 a. As soon as the gas was in- Fig. 3 a–c a) Mass variation of the sample as a function of time and temperature jected (at ≈130 min), a weight loss (≈25 % (in dotted line); b) enlargement of the Fig. 3 a; c) carbon deposition

for Fe2O3; Fig. 3 b) that was associated with in the crucible for the Fe2O3 samples

refractories WORLDFORUM 11 (2019) [2] 85 TECHNOLOGY TRENDS

mixtures. Fe O powders were mixed with ß Carbon 002 -.; 2 3 2 - il! .-.� V. -.l:,,�; - .. - � , .• .,. • • .... .,_ ß Carbon (C) .. ,� . , �"_.:..•1 .. \' - sp • • ,. • ' �..� .,.�-�·, . .. �,� -.....� ·· .. ...J -· � ·--...� · -� sulphur with different amount (1, 5, 10 and ■ .. / ö - Cementite (Fe3 C) ..:- , �. ;-.; ' . ,if ...... ,:� ... f ...... c•� ,. ',•.• • ,�-r- . - -.....�• .� ..� � • ß�� '- � �. -,..,.� ��·��· ·--- - . 30 mass-% S respectively) and with sul- ... �·��ict.�_,.-... ·�� . ,�«•: ��. .. rt4 � • --� , ..." .. r i� phates (1 mass-% BaSO , 1 mass‑% CaSO , ·---��-� ,� ...., ,-:.'1• • ,�·-� -�. 4 4 7 ,.-:.�•L!l.-•��.:�� .. �-� tt' -=�; 1 mass-% MgSO ). Fig. 5 a–b shows TGA ....:i- :J•·· J- A ,� � 4 .",,. , 4�f,...... �:c' �•� .l.t ,.,.,-•'t_,�, •• "".. ,,.➔ -·��.��--•:�•-"' ·.·-�: ��,- results (mass variation as a function in time) ·� ·�.. ; ...• .. �, t-·� � 1��,,\.. · ,;'-�,�•• . �� \ fl . „f�.,__- .-;-�� ,,•I � � and are compared to an experiment with- "' ' -·' •. � ...-c:. � - � � ;:.. � � �-.. ,.._Jl -�G . -·· � • --- ... - . - .... .4' 411!':° .- �-.�, ..... �·,r out sulphur or sulphates additions. • ... Ace.V - Spot Magn Oet WO Exp - •.- •' fll" T .,, - . 0 ..,�o.O kV 5.1 2000x BSE 10.0·· 'r0• .,.·. . 4 FeO <5 µm . _.; ' ...... ,. .. !Jll„t ._ •, ·,:"'..'" - --�..._ • During heating, in inert atmosphere • ••• "'t ·� •. ·, • t t ( . .. . ,.• .•�: ': • i. •. • .•• ..... • , ffl ...... '• I :'_-�� -� • „ I • • • � • . •••• –1 • .,<'.·-,i . . .,...... · (100 ml · min Ar), Fe O + S mixtures \; :. . 2 3 � - . . , �,,,, . . :' ;� . . ... ··.... � .. . .•, ..:, · ...... - . . •· ·.•. . · ,..... ·,. .. ' . , � .�)..•'� .. •. ••• . . " - . .""'· ...... I . have a mass loss associated with sulphur > .•. ,� .•� . .f.i.; .._ •� • ,. . •• :·. . •. •:• . ,... �•· 1 )• • . • l ' -• · �.;, . . ' - • . 1,,. .;,.�.. .. . \ vaporization at 443 °C, proportional to the -..,, :· � ...... • ' ,...:� ... � i1'·...... � � .. \ . •.. . ! . • -.L •.. . .;: . ·.. _· ' . .'-. � •:"' . . ; ·. ·..:. . . ·.- ·. . .-• . ' ,...... ·" . . ... ·.. . . amount of the introduced S (Fig. 5 a). This . •. : �., ...r. r. ·�. .• . . . . . : .. . 1· ' ;,,; • . .- ' � . .. . -t:• !"· ; ' .. � � -�- ,.:.• .. . o!!� ·r ·'-- ... : ·� '·' . :. • . .: . .··,· -�-' .. t.�-� . ..• - . loss of mass is not observed for the mix- . • r... . ·: :w. . .,.·. ., . .. .,,, . ,. . . . , . :,1 .J ;. .-,� . .., •• ... ;, -'-. . "-,.'t� . : • i: . , •-:.· . . •• ' . . ... • . . 'II • • .,. ·. · �'� ,. .,. ; ·◄· • • .\. • tures: Fe O – 1 mass-% BaSO , CaSO , . � •'I. • ·- �,- 60 65 ...... ,.�. . ·..... 2 3 4 4 20 25 30 35 40 45 so 55 70 �· 20 (deg.) . . ··."k•-•�• Ace.V Spot Magn r Oet WO Exp .__...._. MgSO (Fig. 5 b). Indeed, the decompos . · • 20.0 kV 5.0 2500x BSE 10.0 0 4 �

sample Fe2O3 <5 µm; b) after the CO–H2 reaction, the size of particles is reduced 600 °C. As soon as the CO/H2 reducing gas is injected, a weight loss (approximately Complementary measurements: XRD and verse Boudouard reaction catalysed by 25 mass-%) is observed which is due to

SEM observations (Fig. 4) show the quasi the FexOy oxides: a low quantity of sulphur the chemical reduction of the iron oxides

complete transformation of a few milligrams additions prevents the formation of car- by hydrogen. The Fe2O3 sample (reference)

of FexOy in pure cementite phase (Fe3C). The bon [5]. TGA experiments were performed shows a very important carbon deposit. The sp² type carbon is formed and is favoured to quantify the effect of inhibiting sulphur samples containing sulphur and sulphates by the catalyst particles. The catalyst grains on carbon growth rate as a function of the have a very low weight gain that stabilizes are highly fragmented to nanometre-sized S amount in the various solid or gaseous after 200 min. grains independently of the initial grain size, mixtures. due to hydrogen effect and the chemical re- 4.2 Effect of S gas on the amount 4.1 Effect of solid sulphur on the of carbon produced by the duction. During the experiment, the FexOy particles were substantially reduced in size. amount of carbon produced by Boudouard reaction the Boudouard reaction The continuous injection of gaseous sul- 4 Effect of sulphur compounds TGA experiments were performed to val phur has many advantages compared to a on the inhibition of idate the effect of inhibiting sulphur com- solid sulphur addition. It allows providing a the carbon formation pounds on carbon formation, and to quan sustainable amount of sulphur and thus, a This part presents an efficient method to tify the carbon growth rate in function of the more efficient inhibition of carbon forma- inhibit the carbon deposit from the re- S amount and sulphur species in the various tion. The experiments were carried out with

a) b) +50 r----,------, +·o .-----,------r ...... 600 _____ .;..______' 1 - 600 - ! i carbon depo ition 0 ! carb<>n deposition -·-·-· F 0 F '-' e,03 .__, e2 l i i/ Fe, + 1% ßo 400 � 1 400 � i j Fe2 l + 1% Fc1 3 + 1% a • 2 5% 2 «I +r Fep + «l..... ,...._ +25 Fe, , + 1% Mg 0, ,_ -

V, V, «l «I � � -25 -25

- 0 0

400 600 800 0 200 400 600 800 lOOO 1200 Time (min) Timc(min) 0 200 1000 1200

Fig. 5 a–b TGA curves of the samples exposed to the reducing gas mixture at 600 °C:

a) mixture of Fe2O3 powders and pure S (1, 5, 10 and 30 mass-%, respectively);

b) mixture of Fe2O3 powders and sulphates (1 mass-% BaSO4, 1 mass-% CaSO4 and 1 mass-% MgSO4)

86 refractories WORLDFORUM 11 (2019) [2] TECHNOLOGY TRENDS

Fe2O3 at 600 °C under a CO/H2 reducing gas containing SO . +180 2(g) ------t t :- Wi hou 502 +160 ° Fig. 6 shows TGA results (mass variation ' 600 C , , 1 ppm 502 ° - '' 10 C.min l ' / 10ppm SO J , 2 as a function of time) for SO2(g) contents +140 '' J equal to 1, 10, 50, 100 and 1000 ppm, '' - +120 respectively and compared to an experi- 0� ' �� ', C +100 '' ment without SO2(g) addition. Once the gas 0 '' 0 : mixture is injected, all samples underwent ....ro +80 '' ·c � ' ro ,# / a similar mass loss for about 20 min. This is > +60 p.1 ,' (/) qj ,' 50 ppm SO2 the consequence of the Fe O reduction into (/) 2 3 +40 Reducing gas ro F::.f/ ,' pure iron prior to its carburation in iron car- � +20 '' ' ' bide [6]. For the samples exposed to a gas '' mixture containing 1–10 ppm of SO , a 0 '' 2(g) J 100 ppm SO2 '' J dramatic mass gain is observed due to car- -20 '' l000ppm SO2 bon formation; roughly like what could be 0 30 60 90 120 150 180 210 240 found without SO . When 50 ppm of SO 2(g) 2 Time (min) was injected, a lower mass gain is observed, Fig. 6 TGA curves of the Fe O samples exposed to the reducing gas mixture containing and the carbon growth rate has been de- 2 3 small fractions of SO at 600 °C (continuous curves) creased by a factor 5 in comparison with 2 the experiment without SO2. For 100 ppm and 1000 ppm of SO2, the carbon growth spectra of the carbon formed in the two size reduction. Therefore, poly-aromatic lay- rate values have been decreased by a fac- types of gas are totally different. G band ers are formed around the catalyst particles tor 70. However, a slightly higher mass gain is observed in the carbon formed under in the form of encapsulating shells. These is observed for the sample with 100 ppm of 100 % CO, and D and G bands are ob- layers are sufficiently large with regards to

SO2 in comparison with 1000 ppm. served when the gas is a mixture of CO the laser diameter which explains the no

and H2. activation of the defect-induced D band in 4.3 Structural evolution of the In CO atmosphere, well organised sp2 car- the corresponding Raman spectrum [6]. In carbon bon is formed around the catalyst particles CO + H2 atmosphere, the corresponding Ra- and the defect-induced D bands are not man spectrum exhibits the defect induced D 4.3.1 Raman spectroscopy observed. Only chemical reduction of the band suggesting disordered carbon. Cata- Raman spectroscopy is an ideal technique catalyst particle occurs without significant lyst particles are chemically reduced and to characterize in situ sp2 carbon materials. It is the most sensitive to explore the full range of the structural states from perfectly crystalline to amorphous [7]. The typical Raman spectrum of sp² carbon is reported in Fig. 7 c and shows three prom inent features. The so-called G band located at approximately 1580 cm–1 corresponds to the in-plane bond stretching of carbon atoms (E2g symmetry). When defects are 200 nm present, the crystal symmetry breaks down and additional bands appear at specific frequencies depending on the excitation energy used. These bands are called the defect bands: D (≈1250–1400 cm–1) and D’ G G –1 (≈1600–1630 cm ). At relatively low de- D fect density, the intensity ratio of the defect- induced D band and the symmetry-allowed ) 1000 1500 ZOOO 1000 2000 G band increases with disorder. Wavenumber(c nr 1) Pure iron and iron oxides powders were submitted for 30 min at 600 °C under re- Fig. 7 a–c Schematic representation of the interaction of the Raman excitation laser and ducing atmosphere: 100 % CO and CO/H2 the carbon particles in shell around catalyst particles in 100 % CO and in nano-fibres mixture. in CO + H2 atmosphere. TEM micrographs of nano-carbon shells and filaments formed Fig. 7 summarizes the results obtained us- à 600 °C in: a) 100 % CO; and b) 70,0 % CO, 3,0 % CO2, 10,0 % H2 and 17,0 % N2 from ing Raman spectroscopy and TEM. Raman Fe2O3 nanometer size catalyst refractories WORLDFORUM 11 (2019) [2] 87 TECHNOLOGY TRENDS

100 100 90 A- 100% CO 90 80 80 70 70 §:60 §:50 � � I 50 ◄ � I 50 LLs 40 LLs 40 30 30 20 ti 20 10 10 • B- 95% CO + 5% H2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 2.0 2.5 3.0 lilG 100...... ------.-----,.------T"----.---..------. 100 90 C- 90% CO+ 10% H2 90 D- lndustrial gas ...... • ,t "' 80 80 ► � : : ♦ ...... ►+ 70 "' . 70 �� ► . . • ., . ► -- ► 1..• •· ◄ "'J/1,,.� ...· . '<')., A_•/ ... �60 ◄- , .. -50 � ·�·•1...i+ � ... � �, 1,�.. .., : ► • ' • ◄ .... ♦ ► ♦ 50 ◄ I 50 • . � ,► ...... -·· • • • "' ◄ ... "' ♦ LL 40 "' LL 40 "'• .• •• • ... 30 •• • 30 20 • 20 • 10 10 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 lilG lilG

• Fe < 5µm ◄ Fe203 35 nm • <5µm • FeO < 5µm Fe203 "' Fe O < 5µm • Fe 0 160-250µ m 3 4 2 3 • Natural graphite ► Fe203 1000-2000 µm

Fig. 8 Evolution of G band width at half-maximum (FWHM(G)) versus the intensity ration of the D relative to G band of series of in situ measurement of carbon formed under different iron catalyst at different gases at 600 °C (except for C at 500 °C)

highly fragmented. Therefore, small-sized closed to the bottom part of the diagram 4.2.2 Microscopic characterisation of samples carbon nanofibres are formed exhibiting a with a moderate FWHM(G), due to the tem- large amount of poly-aromatic boundaries. perature of the experiment. As mentioned To better understand the effect of sulphur This eventually leads to the activation of the above, this result is in accordance with the on the inhibition of the carbon produced defect induced D band without the local lack of D band on the Raman spectra, due by the reverse Boudouard reaction cata-

structural order being necessarily disturbed to the morphology of carbon in onion. The lysed by Fe2O3, the samples were observed [6]. addition of 5 % of hydrogen in the gas after TGA experiments, by Scanning Elec-

Fig. 8 reported the FWHM(G) versus ID/IG mixture induce a moved to the right part tron Microscopy (SEM) and High-Reso- for all the samples at each gas. According of the diagram (Fig. 8 b), with an important lution Transmission Electron Microscopy

to [8], this graphic show the evolution of increase of the FWHM(G). The increase of the (HRTEM).

the “crystallite size and defect” of the car- H2 content favoured an increase of the ID/IG TEM micrographs of the samples, not ex-

bon. The most well crystalline sp² carbon ration and a relative staggering of the point posed to SO2, show the presence of small is natural graphite, and it plotted closed (Fig. 8 c). This spreading should be moder- black iron carbide particles with a reduced to the origin of this graph (orange circle). ated by the low quality of Raman spectra size compared to the initial iron oxide grain The less organised carbon could be show obtained for this test. The experiment using size (Fig. 9). A part of the carbon forms

ideally in the top of this diagram (Fig. 8) an analog of industrial gas (Fig. 8 d) with nanofibres growing from Fe3C particles. –1 with FWHM(G) upper to ≈70 cm . Fig. 8 a a CO/H2 ratio closed to the gas mixture C, These carbon fibres are composed of poly shows the results for the carbon obtain shows results similar to the data obtains in aromatic layers surrounding an amorphous to a 100 % CO gas. All the data are very Fig. 8 c. core (Fig. 9 c–d).

88 refractories WORLDFORUM 11 (2019) [2] TECHNOLOGY TRENDS

The rounded shape of these particles may suggest a molten state. This observation was previously observed between 800– 1000 °C [9, 10], and explained by Krause and Pötschke [1]: In the binary phase diagram Fe–C, the eutectic melting point drop from 1153 °C to temperature lower than 500 °C, and at this temperature droplets of carbon bearing iron are visible.

The sample exposed to 1000 ppm of SO2(g) exhibits a quite different morphology. EDS analysis reveals only the presence of iron carbide particles such as Fe3C and/or Fe5C2 according to XRD. Considering their sizes, those particles do not seem to have been fragmented or fragmented to a lesser extent. A TEM micrograph for a sample exposed to

1000 ppm of SO2(g) is presented in Fig. 10 a. No filamentous carbon is observed on the micrograph and the iron carbide particles are systematically covered by a nanometer- sized thick shell. High resolution STEM- HAADF images show a two-layered shell (Fig. 10 b–c) The inner layer is very thin (0,5–1 nm). This inner layer is rich in sulphur with a rather high content of iron, which suggests the formation of an FeS iron sulphide.

5 Discussion about the Fig. 9 a–d TEM micrograph of nano-carbon filaments for different catalysts formed at 600 °C in: a) 71 % CO, 3 % CO , 11 % H and 15 % N a Fe O 35 nm; b) Fe O 1–2 mm; mechanisms of carbon formation 2 2 2 2 3 2 3 c) Fe O <5 μm; and d) FeO 2–5 μm and carbon inhibition by sulphur 3 4 Under 100 % CO atmosphere, the carbon formation by the Boudouard reaction is a well know process. The iron oxides are reduced by CO and converted to iron carbide

(cementite). Liquid iron/Fe3C nanometric droplets are formed which dissolve carbon until the saturation limit. Carbon is segre- gated on the liquid surface droplet [1]. The catalytic deposit of carbon originates from a dissolution-precipitation mechanism of carbon. The carbon has a high structural organisation and forms polyaromatic shells around big catalytic oxide particles. When hydrogen is present, the reactions become more efficient. The phase trans formations (Fe2O3 → FeO → Fe → Fe3C) result in a dramatic increase in the specific surface areas of catalytic particles. H2 favours the nucleation and growth of carbon fibres on the iron-carbide particles. Fig. 11 reports the different steps of the for- Fig. 10 a–c Micrographs of the sample after TGA experiments exposed to mation of carbon nanofibers by the reverse 1000 ppm of SO2(g): a) TEM image representative of the sample; b),

Boudouard reaction in CO–H2 atmosphere: c) high resolution HAADF-STEM images of a particle surrounded by a two-layer shell refractories WORLDFORUM 11 (2019) [2] 89 TECHNOLOGY TRENDS

• Step 1: reduction by H2 and CO with for-

mation of suboxides of iron (Fe2O3 → FeO → Fe) recrystallization of particles and decrease of iron oxide size to nanometer- sized particles;

• Step 2: liquid iron/ Fe3C nanometric droplets; • Step 3: grow of polyaromatic carbon nanofibres. The mechanism of the carbon formation is inhibited by sulphur or sulphates additions. This result is confirmed by 20 h experiments in laboratory. On the surface of the iron ox-

ide particles, FeS or FeS2 iron sulphides are formed (Fig. 12). The formation of these sulphides can be partial or total depending on the size of the iron oxide particles, the heating time and the reaction time. Sulphur does not prevent the reduction of

Fe2O3 by H2 and CO and the fragmentation

of the Fe2O3 grains. Nevertheless, it limits the formation of sp2 carbon. Sulphur diffuse into the crystallite structure or through the Fig. 11 Mechanism of carbon nanofibres formation by the reverse Boudouard reaction in grain boundaries. Moreover, it is not neces-

CO–H2 atmosphere sary to transform all the iron into iron sulphides to inhibit the carbon deposition by the reverse Boudouard reaction. However, the FeS layer is not stable, S ad- sorption is a dynamic phenomenon and there are always some defects in the FeS layer which will lead to a slow continued carbon formation. Fig. 13 reports the mechanism of carbon

inhibition in CO–H2 atmosphere. 6 Implication for industrial processes Fig. 12 SEM micrograph of iron sulphides at the surface of catalytic iron particles The carbon formation from the Boudouard reaction and the nanocarbon formation were extensively studied for the last two decades. Nowadays, this phenomenon appears to be highly important for the development of new high temperature industrial processes. For example, in new energy field, the biomass and the waste gasification or the low C blast furnaces are highly vulner- able when the Boudouard reaction occurs. The refractory lining of these reactors could

be highly damaged. Moreover, CO/H2 gas will be efficiently used for reducing the power cost of industrial processes, and a technical solution should be developed for preserved refractory materials. Indeed, all these new developments are in relation with the global objective of reducing green-

Fig. 13 Inhibition of carbon formation by S in CO–H2 atmosphere house gas emissions. The injection of H2 gas

90 refractories WORLDFORUM 11 (2019) [2] TECHNOLOGY TRENDS

in CO mixture in refractory devices devel- CO + H2 reducing atmosphere (composition formed polyaromatic shells around rela- oped new challenges with the limitation of R). Reference powders of pure iron oxide tively big catalytic particles. the damage of ceramics material. This de- were also tested under the same experi- The addition of H2 in CO favours the forma- struction was well controlled in the case of mental conditions. tion of polyaromatic carbon with nanofibre only CO by empirical knowledge of refrac- The castable C1 (with 5 mass-% Fe2O3) was morphology. This morphology of carbon tories companies, however H2 favoured the destroyed by a “bursting reaction”. A sig- could be the cause of the destruction of

Boudouard reaction and eventually the for- nificant carbon deposit in the cementitious refractory materials test in CO + H2 atmos- mation of a specific sp2 carbon as demon- matrix was observed and analysed by Ra- phere. Laboratory experimentations showed strated in the paper. The problem is induced man spectroscopy. The castables C2 and C3 that sulphur additives inhibited the CO dis- by the formation of carbon nanofibres in an containing S and BaSO4 additives did not sociation and the formation of carbon. The abundant manner linked to the presence of exhibit significant carbon deposit and were practical conclusions of this study will be

CO + H2 in the future industrial atmosphere. not damaged by the reverse Boudouard applied to develop new CO/H2 resistant re- Consequently, this study is helpful for the reaction (Fig. 14). These results were con- fractories. development of laboratory and industrial firmed by Thermogravimetric Analyses solutions, for example by modification of (TGA), under reducing atmosphere: com- References carbon morphology. position R: 71 % CO, 3 % CO2, 11 % H2, [1] Krause, O.; Pötschke, J.: The predictability of CO

The benefit effect of S to inhibit the for- 15 % N2. resistance of refractory materials by state-of- mation of carbon was applied to develop However, the effect of S additives to inhibit the-art test methods – a technical and scien-

CO/‌H2 resistant refractories. Laboratory the formation of carbon has limits: the sul- tific approach. Interceram 57 (2008) 176–180 experiments were made using a commer- phur additives become ineffective beyond [2] Bost, N.; et al.: The catalytic effect of iron ox- cial castable. The chemical composition 900 °C. In some industrial applications, the ides on the formation of nanocarbon by the [mass‑%] of the castable is: refractories are subjected to cyclic condi- Boudouard reaction. J. Europ. Ceram. Soc. 36 tions: oxidising and reducing. Under these (2016) 2133–2142 conditions, the stability of FeS and FeS is [3] Walker, P.L.; Raicszawski, J.F.; Imperial, G.R.: Al2O3 SiO2 CaO Fe2O3 2 Carbon formation from carbon monoxide- 81 16 2,5 0,6 not guaranteed. An addition of sulphurous gas (100 ppm) in hydrogen mixtures over iron catalysts. J. Phys.

CO/H2 gas should be an effective solution to Chem. 63 (1959) 140–149 To promote the formation of carbon and limit the carbon formation and the damage [4] Turkdogan, E.T.; Vinters, J.V.: Catalytic effect of to increase the degradation of refractories, of refractories. iron on decomposition of carbon monoxide: I.

5 mass-% Fe2O3 nanoparticles were mixed. carbon deposition in H2–CO mixtures. Metall. Three different formula were prepared, and 7 Conclusion Trans. 5 (1974) 11–19 a low quantity of sulphur was incorporated The carbon deposit in refractories is cata- [5] Berry, T.F.; Ames, R.N.; Snow, R.B.: Influence of in two samples: lysed by iron and iron oxides. This reaction impurities and role of iron carbides in depos

• C1: castable + 5 mass-% Fe2O3; is highly favoured by the presence of H2. The ition of carbon from carbon monoxide. J. Amer.

• C2: castable + 5 mass-% Fe2O3 and mechanism of the catalytic decomposition Ceram. Soc. 39 (1956) 308–318 0,5 mass-% solid sulphur; of carbon monoxide via the Boudouard in [6] Bost, N.; et al.: Probing the structural organiza- 2 • C3: castable + 5 mass-% Fe2O3 and the presence of H2 is better understood. tion of sp carbons obtained by the Boudouard

3,3 mass-% BaSO4. Two different carbons are formed in specific reaction using in situ Raman scattering in re- After curing, the three castables were pre- gas composition. In 100 % CO, the sp2 car- ducing conditions. Vibrational Spectroscopy 87 fired in air at 900 °C, for 8 h. Then, they bon has a high structural organisation with (2016) 157–163 were submitted, for 5 h, at 600 °C under important coherent domain diameter and [7] Ammar, M.R.; Rouzaud, J.N.: How to obtain a reliable structural characterization of polished graphitized carbons by Raman microspec- troscopy. J. Raman Spectroscopy 43 (2011) 207–221 [8] Ammar, M.R.; et al.: Characterizing various types of defects in nuclear graphite using Ra- man scattering: heat treatment, ion irradiation and polishing. Carbon 95 (2015) 364–373 [9] Yazdanbakhsh, A.; et al.: Challenges and benefits of utilizing carbon nanofilaments in cementitious materials. J. of Nanomaterials 2012 (2012) 2012, Article ID 371927, 1–8 [10] Drzal, L.; Fukushima, H.: Conference proceed- Fig. 14 Carbon deposit and destruction of C1 (a); ings of high performance fillers, paper 15, effect of sulphur on C2 and C3 castables after exposure to reducing gas (b) 2006 refractories WORLDFORUM 11 (2019) [2] 91