- Email: [email protected]
Innovative carbon-bonded filters based on a new environmental-friendly binder system for steel melt filtration
Accepted Manuscript Title: Innovative carbon-bonded filters based on a new environmental-friendly binder system for steel melt filtration Authors: C. Himcinschi, C. Biermann, E. Storti, B. Dietrich, G. Wolf, J. Kortus, C.G. Aneziris PII: DOI: Reference:
S0955-2219(18)30526-0 https://doi.org/10.1016/j.jeurceramsoc.2018.08.029 JECS 12054
To appear in:
Journal of the European Ceramic Society
Received date: Revised date: Accepted date:
14-5-2018 14-8-2018 20-8-2018
Please cite this article as: Himcinschi C, Biermann C, Storti E, Dietrich B, Wolf G, Kortus J, Aneziris CG, Innovative carbon-bonded filters based on a new environmentalfriendly binder system for steel melt filtration, Journal of the European Ceramic Society (2018), https://doi.org/10.1016/j.jeurceramsoc.2018.08.029 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Innovative carbon-bonded filters based on a new environmental-friendly binder system for steel melt filtration
a
IP T
C. Himcinschia,*, C. Biermannb, E. Stortib, B. Dietrichc, G. Wolfc, J. Kortusa, C.G. Anezirisb
Institute of Theoretical Physics, TU Bergakademie Freiberg, Leipziger Straße 23, 09599 Freiberg,
b
SC R
Germany
Institute of Ceramic, Glass and Construction Materials, TU Bergakademie Freiberg, Agricolastraße
17, 09599 Freiberg, Germany c
U
Foundry Institute, TU Bergakademie Freiberg, Bernhard-von-Cotta-Straße 4, 09599 Freiberg,
M
ED
* [email protected]
A
N
Germany
PT
Abstract:
New carbon-bonded alumina filters for steel melt filtration were developed. The carbonaceous
CC E
matrix was based on a new, environmental friendly binder system based on lactose and tannin. The filter preparation was analogous to the production of conventional foam filters according to the Schwartzwalder process. The processing as well as the rheology of the slurries was investigated. An
A
addition of n-Si increased the carbon yield and the cold crushing strength (CCS) of the samples. Higher values of CCS were obtained after coating of the filters with alumina. The material was characterized by scanning electron microscopy, X-ray diffraction and Raman spectroscopy. The applicability of these new filters was assessed in impingement tests with a steel melt, in which three out of four recipes survived the thermal shock. 1
Keywords: Steel filtration; Lactose; Tannin; Ceramic foam filters; Raman Spectroscopy
1. Introduction
IP T
Carbon-bonded ceramic foam filters are established for applications in steel melt filtration, to
remove non-metallic inclusions from the molten steel at temperatures between 1400 and 1650°C,
SC R
and to guarantee the reliability required by the customer, especially for safety-relevant components[1].
U
The state of art is the use of novolac or resoles as binder[2]–[5]. Pitch binders have the disadvantage of
N
releasing harmful components such as carcinogenic products (benzo[a]pyrene) during production
A
and operation. An improvement was achieved by the application of Carbores®P (CARBOnaceous
M
RESin), which is a kind of high melting artificial coal-tar resin. The yield of benzo[a]pyrene for Carbores is much lower than for traditional pitches, 50000 vs < 300 ppm. However, the limit in
ED
Germany is fixed at 50 ppm[6]. Thus, environmental friendly resins are generally preferred. Recently, a new environmental friendly alternative binder system for carbon-bonded MgO-C materials was
PT
presented[7]. In addition, a patent describing it was applied[8]. The structure of this new binder
CC E
system, based on lactose and tannin, is similar to the novolac structure. During the curing step, a resit network structure comparable to the novolac lattice is formed. The stability of the cured state is also similar[7]. In the residual carbon, the structure of the resit network is an isotropic glassy phase.
A
This isotropic structure leads to a lower oxidation and corrosion resistance and worse thermomechanical properties compared to more ordered carbon forms such as graphite[9][10][11][12]. A way to enhance the quality of carbon-bonded refractory products is the addition of antioxidants[13], which hinder carbon oxidation. During processing, metallic antioxidants, e.g. Al are added to the raw materials[13]–[19]. This addition results in the improvement of oxidation resistance and of the mechanical strength[18][20]–[22]. 2
Yamaguchi et al. [23]determined the effect of certain oxides on the oxidation resistance of graphite and amorphous carbon. For instance, TiO2 prevents the oxidation of graphite by donating electrons and therefore stabilizing its structure. The new environmental friendly binder system is a combination of several raw materials and additives[7]: hexa for hardening similar to novolac binders, n-Si for increasing the carbon yield[24][25],
IP T
TiO2 as electron donator [13][24] and Al as antioxidant [13]. A mixture of lactose and tannin showed a large volume expansion during heat treatment and the resulting product was extremely porous.
SC R
Thus, SiO2 was added as antifoam agent [26]. The average carbon yield of the new binder system after treatment in reducing atmosphere at a temperature of 1000 °C was 51.5 ± 0.5 %, which is similar to
U
the yields of the conventional novolac binders[7].
N
Open-cells foams are the most widely used structures in the case of metal melt filtration[27].
A
However, so far only bulk MgO-bricks were produced on basis of the new binder system.
techniques was given by Luyten et al.[28].
M
Several methods for production of foam structures were developed. An overview of the different
ED
In the present work, ceramic foams based on the new environmental friendly binder system were manufactured according to the replica technique patented by Schwartzwalder in 1963[29].
PT
Polyurethane foams were used as base for the replica technique. The porosity (defined as pores per
CC E
inch, ppi) of foam filters for metal filtration is always chosen based on the metal properties. For example, foams with 10 ppi and 30 ppi are commonly used for steel and for aluminum melt filtration, respectively[27]. Thus, in our case, 10 ppi foams were used. After the impregnation step, a thermal
A
treatment was applied in order to cure the binder and decompose the polyurethane foam. After treatment, the material consists of a carbonaceous matrix surrounding the ceramic particles. The foams possess an open-cell structure with cells of pentagon-dodecahedronal geometry linked by triconcave struts[27]. Moreover, active oxide coatings were applied to some of the filters in order to increase the cold crushing strength and to improve the collection of oxide particles from the steel
3
melt during operation[30]. The foam filters were characterized in terms of structure and mechanical properties and tested in impingement tests in contact with steel melt. The structure of the lactose-tannin-based binder system and in particular the effect of n-Si was studied by Raman spectroscopy. Raman spectroscopy is a very sensitive detection method for specific vibrational fingerprints of distinct atomic groups in molecules, crystalline or amorphous
IP T
systems. Thus, it is suitable for the identification of the chemical structure and / or changes in the chemical structure. Nowadays, Raman spectroscopy is widely recognized as a non-destructive
SC R
standard method for the characterization of carbon-containing materials because of the "fingerprint"
U
and information on bonding that it can deliver for such systems [31]–[34].
N
2. Materials and characterization techniques
A
The raw materials used for the preparation of the investigated Al2O3-C foams were aluminum oxide
M
(Martoxid MR70, Martinswerk, Germany, 99.80 wt.% Al2O3, d90 ≤ 3.0 µm), lactose (α-Lactose-
ED
monohydrat DAB 9, DM-Markt, Germany) and tannin (Quebrachoextrakt, type Indusol ATO., Otto Dille, Germany) used both as binders and carbon sources[7], hexamethylenetetramine (Momentive,
PT
Germany), fine natural graphite (AF 96/97, Graphit Kropfmühl, Germany, 96.7 wt.% carbon, 99.8 wt.% < 40 µm), and carbon black powder (Luvomaxx N-991, Lehmann & Voss &Co., Germany, ≥99.0
CC E
wt.% carbon, >0.01 wt.% ash content, primary particle size: 200-500 nm). The following additives were used: ammonium-ligninsulfonate C12C (Otto Dille, Germany) as temporary binder and wetting agent, Castament VP 95 L (BASF, Germany) as dispersing agent, Contraspum K 1012 (Zschimmer &
A
Schwarz, Germany) as antifoam agent[35]; TiO2 (nano-powder TR, ≥ 99 wt.% rutile, Crenox GmbH, Germany) and aluminum metal (Al-Grieß – fraction 0-75 µm, ≥ 99.6 % Al, d50 = 29.0 µm, Hoesch Granules GmbH, Germany) as antixoxidant agents and SiO2 (RW-Filler weiß, RW silicium GmbH, Germany) as antifoam agent[7]. To increase the carbon yield, P-doped n-Si (grade 1a, > 99.9 % Si, particle size < 63 µm, Silchem, Germany) was used in one batch. 4
The preparation of the carbon-bonded filters was divided into two steps according to the procedure described by Emmel et al.[27]. In the first step, a highly viscous slurry for impregnation of the PU foam was prepared in a Hobbart-type mixer (ToniTechnik, Germany). Here, the powder components listed in Table 1 were weighted and dry mixed for 5 min. Then, the liquid additives (Castament VP 95 L, Contraspum K 1012) were dispersed in deionized water and added to the dry mixture. Further
IP T
deionized water was added step by step until a plastic mass was obtained. This mass was mixed for 5 min for best homogenization. Finally, the rest of deionized water was added to get an impregnating
SC R
slurry with the total solid content of 78 wt.%.
Table 1: Composition of the impregnating and spraying slurries without (recipe 1: AC5-2AH) and with
U
n-Si (recipe 2: AC5-2AHSi), and of the alumina coating slurry.
Coating material
N
Filter material Weight fraction (%)
Al2O3 Martoxid MR 70
66.0
lactose/tannin=5/1
20.0
-
Carbon black N-991
6.3
-
7.7
-
M
ED
100
Mass in wt.% (related to total solid content)
Hexametilenetetramine
10.0 b
-
TiO2
0.5
-
Al
0.1
-
A
CC E
Additives
PT
Graphite AF 96/97
A
Material
SiO2
4.0
-
n-Si (recipe 2)
5.0 b
-
Castament VP 95 L
0.3
0.3
Contraspum K 1012
0.1
-
5
Ammonium-ligninsulfonate C12C
1.5
1.5
Total solid content (wt.%) Impregnating slurry
78.0
-
Spraying slurry
70.0
65.0
b
related only to binder content of 20 wt.%
IP T
This slurry was used for the impregnation of PU-foam blocks with dimensions of 50 mm x 50 mm x 22 mm. For this step the slurry was pressed by hand into the foams to guarantee a complete coating of
SC R
the PU framework. To remove the excess slurry, the foam blocks were then pressed through a special roll-pressing device with two counter-rotating rolls. The gap between the two rolls was in our case 4
U
mm. Subsequently, the wet impregnated foams were dried at room temperature to a constant
N
weight. A further processing step was necessary to obtain foams with a homogenous and sufficient
A
thickness of ceramic layer. Thus, a spraying slurry with a total solid content of 70 wt.% was prepared as described above. This slurry was applied by spraying by means of an airgun HS-25HVLP
M
(Krautzberger, Germany) at a spraying pressure of 3 bar in a spraying chamber. The distance between
ED
the spray gun (nozzle) and the foam was kept at approximately 10 cm. The four smaller sides were sprayed for about 2-3 s and the two larger sides for 5-6 s. The weight of each foam at the end of the
PT
spraying step was about 27 g in the wet condition. Afterwards, the samples were dried at room
CC E
temperature for 24 h.
The dried foams were hardened inside a drying chamber up to a temperature of 180°C according to the schedule in Fig. 1. This treatment was followed by pyrolysis in a retort filled with calcined petcoke
A
(Müco, Germany; particle size 0.2 – 2 mm), in order to approach reducing conditions. The heating rate was 1 °C/min up to 1000°C and 180 min holding time. In addition, every 100 °C a holding step of 30 min was included to prevent damage from volatile release. After the pyrolysis process the samples cooled down to room temperature inside the retort in the oven. The final weight of the carbonbonded filters was approximately 18 g.
6
b)
SC R
IP T
a)
Fig.1: Heating procedure for hardening (a), pyrolysis of the filters (b, blue line) and sintering of the
U
extra coating (b, orange line)
N
Furthermore, some coked samples were spray coated with a functional coating based on pure
A
alumina[30][36]. The composition of this slurry is presented in Table 1. The coating procedure was
M
similar to the previous one, described above. In this case, the weight of the wet filters reached about 29 g. The second coating was also dried at room temperature. Afterwards, the samples were treated
ED
again inside a retort filled with coke. The heating rate was again 1 °C/min up to 300 °C with 30 min
of the coating.
PT
holding time; 3 °C/min up to 1400 °C and 5 h dwell time, to promote sintering of the alumina grains
CC E
In the end four different recipes were produced, i.e. (i) without n-Si and Al2O3 coating (AC5-2AH) and (ii) with Al2O3 coating (AC5-2AH-alox) as well as (iii) with n-Si without Al2O3 coating (AC5-2AHSi) and (iv) with Al2O3 coating (AC5-2AHSi-alox), respectively, see Fig. 2.
A
The rheological behavior of the slurries was measured with the aid of a rheometer RheoStress RS 150 (Haake, Germany). For the measurement of the flow curves, coaxial cylinder measuring systems of the type Z38 and Z40 DIN were used for the impregnating and spraying slurries, respectively. The flow behavior was determined by measuring the viscosity and shear stress under variation of the
7
shear rate. This was changed linearly from 0.1 to 500 s-1 (up to 1000 for the spraying slurries), and
PT
ED
M
A
N
U
SC R
IP T
kept at this level for 90 s, before decreasing it at the same rate down to the initial 0.1 s-1 value.
CC E
Fig. 2: Carbon-bonded Al2O3-C filters without (A, B) and with (C, D) n-Si; without Al2O3 coating (A, C) and with Al2O3 coating (B, D)
A
The microstructural characterization of the filters after heat treatment was carried out by means of a digital light microscope (VHX-200 D, Keyence, Japan) and a scanning electron microscope, equipped with an EDS-detector (SEM XL30, Philips, Netherlands). In order to determine the thickness of the sprayed coating, broken struts were investigated in cross section. The open porosity was evaluated using a mercury intrusion porosimeter AutoPore V 9600 from Micromeritics Instrument Corporation (USA) and the software Micro Active. 8
The X-ray diffraction analysis was carried out on a standard diffractometer (PANalytical X’Pert Pro MPD 3040/60, Almelo, Netherlands). The instrument operates in Bragg-Brentano geometry with a fixed divergence slit and rotating sample stage. Standard Cu Kα radiation and a scan step size of 0.013° with 30 s holding time per step were used. The X-ray source was operated at 40 kV and 40 mA. The analyses were performed on ground filter struts using the back loading technique to
IP T
minimize preferred orientation. The cold crushing strength (CCS) was determined on a universal testing machine (TT 2420, TIRA
SC R
GmbH, Germany) using a 1 kN load cell. The displacement rate was 3 mm/min up to a force of 5 N. At this load the loading rate was changed to 1 MPa/s. The experiment was terminated when a strength loss of 80 % was reached. From each batch, five specimens were tested in order to obtain average
N
U
values and standard deviations.
A
The specimens for the Raman spectroscopy measurements were produced from a slurry without any
M
graphite and carbon black as well as without Al, TiO2 and n-Si. This slurry was cast into 30 mm x 30 mm square molds and dried at room temperature. Afterwards, two other slurries (one with TiO2 and
ED
a second with n-Si) were prepared. These slurries were cast in a thin layer on top of the dried previously cast substrates. A second drying step was also needed. Thus, several different specimens
PT
were prepared, i.e.:
CC E
(i) Al2O3/tannin/lactose/hexa/SiO2: with different thermal treatments (curing at 180 °C and coking at 200 °C, 300 °C, 400 °C and 500 °C) and (ii) Al2O3/tannin/lactose/hexa/SiO2/n-Si: with different thermal treatments (curing at 180 °C and
A
coking at 200 °C, 300 °C, 400 °C and 500 °C), respectively. The Raman measurements were performed using a Horiba Jobin Yvon Labram spectrometer equipped with a 600 mm-1 grating and a Peltier cooled CCD detector. The measurements were done in backscattering geometry. For excitation the 532 nm line of a frequency-doubled Nd-YAG laser was employed. In order to avoid the heating of the samples by the laser, a low laser power of 0.4 mW 9
was used. The light was guided and focused through a microscope objective with 20X magnification (ca. 10 µm focus diameter), which ensured a lower power density in comparison to microscope objectives with higher magnification and lower focus diameters. This ensured a negligible influence of the laser heating, except for the samples annealed at 200°C, for which a colour change was visible on the measured spot.
IP T
Finally, for the impingement tests, three filters of each batch were embedded in furan-bonded sand molds. Approximately 9 kg of steel were molten and directly poured on the samples, without
SC R
preheating them. The steel was always freshly molten before each filter impingement.
U
3. Results and discussion
A
N
3.1. Rheological behavior
M
In Figs. 3a and 3b the dynamic viscosity of the impregnating and spraying slurries as a function of the shear rate are shown. For comparison, the curves from the reference system (with pitch-based
ED
binder) are also included. A clear shear thinning behavior was observed for all slurries. As expected, the impregnating slurries exhibited a much higher dynamic viscosity than the spraying counterparts,
PT
as is it required for a proper impregnation process. The measurement for the AC5 reference had to be interrupted at 350 s-1 because of the limited shear achievable by the instrument. The same limit
CC E
was not reached by our slips since the reference impregnating slurry had a higher solid content (about 81.6 % versus 78 %). The rheological behavior at the investigated shear rates was very similar
A
to the reference, with dynamic viscosity around 3000 mPa·s at 300 s-1. The spraying slurries were ideal for the coating step, with a dynamic viscosity of about 100 mPa·s at 1000 s-1. Such behavior resulted in a very good sprayability, and at the same time in a good adhesion to the PU foam skeleton after the impact. Overall, no remarkable difference was observed when comparing the formulations without (AC5-2AH) and with n-Si (AC5-2AHSi), with the AC5-2AH sample showing only slightly higher viscosity. Despite having similar solids content, the reference spraying slurry was less 10
viscous than both samples prepared with tannin and lactose. This effect was likely due to the powder additives present in the new formulations, silica in particular.
N
U
SC R
IP T
a)
A
CC E
PT
ED
M
A
b)
Fig. 3: Rheology of slurries AC5-2AH and AC5-2AHSi for impregnation (a) and spraying (b), respectively.
11
3.2 Microstructure Optical micrographs of the carbon-bonded samples (without coating) are presented in Fig.4. The reference filter produced with the pitch binder showed thicker struts than the AC5-2AH and AC52AHSi. In addition, a large number of cracks were observed in particular on the AC5-2AH sample, which explain the lower cold crushing strength of this batch (see below). The addition of silicon
IP T
increased the carbon yield of the binder system, possibly limiting the formation of cracks in the AC52AH samples. Nevertheless, in order to obtain foams with structure similar to the reference one (i.e.
SC R
almost crack-free), it may be beneficial to extend the spraying process. Since the environmental
friendly binder system only yields 51.5 wt.% carbon against 80 wt.% of Carbores P, the filters will be sprayed to a slightly larger weight (i.e. 2-3 g in the wet condition) in future tests. Light blue particles
U
in the macropores of the AC5-2AHSi sample are coke residues, which probably interacted with some
A
N
silicon from the filter during the heat treatment at 1400 °C (metallic Si melts at 1414 °C).
M
In SEM investigations, broken struts were characterized. Fig. 5 shows characteristic struts of the four batches. In the center of the struts, large cavities left from the degradation of the initial PU foams
ED
with the typical triangular shape were present. There was no visible boundary between the material from the impregnating and from the spraying slurry. Large pores within the Al2O3-C filter struts were
PT
likely generated during heat treatment.
CC E
In Figs. 5b and 5d the Al2O3 coating is clearly visible as a homogenous and dense layer of material over the porous substrate. In fact, only the coating material was able to sinter during the heating treatment at 1400 °C, while the presence of carbon in the base material hindered this process. The
A
thickness of the alumina coating layer varied from about 50 µm to 100 µm.
12
IP T M
A
N
U
SC R
a)
A
CC E
PT
ED
b)
c) Fig. 4: Optical micrographs of (a) the reference filter (AC5), (b) AC5-2AH filter and (c) AC5-2AHSi, respectively.
13
b)
c)
d)
U
SC R
IP T
a)
A
M
coating (a, c) and with Al2O3 coating (b, d)
N
Fig. 5: Microstructure of the coked Al2O3-C filters without (a, b) and with (c, d) n-Si; without Al2O3
3.3 Open porosity
ED
The binder content of the new environmental friendly binder system was comparable to the conventional foam filters at 20 wt.%. However, due to the different evolution of volatiles during
PT
pyrolysis, we expected slightly different final porosities. The open porosity of the uncoated samples
CC E
of the lactose-tannin binder system was 50.8 % for AC5-2AH and 54.3 % for AC5-2AHSi. The conventional filters with 20 wt.% Carbores®P usually show an open porosity of about 30 %[27][35]. Thus, the new binder system yielded a much higher open porosity compared to traditional pitch
A
binder. Nevertheless, the filters could be easily handled after the heat treatment. 3.4 Cold crushing strength Table 3 gives the mean values and the standard deviations of the CCS of the four batches of filters and of the reference (AC5, bonded with Carbores P pitch). The CCS of the uncoated, Si-free batch was the lowest at 0.05 MPa, which was still enough to manipulate the filters but much lower than for the 14
reference samples. The use of n-Si as additive in the new binder system showed a positive influence on the cold crushing strength, which was nearly doubled. It should be pointed out, however, that these filters were slightly heavier than the ones without silicon, and that both batches prepared with the new binder system showed a lower weight than the reference samples. As reported by Emmel et al., the crushing strength of ceramic foam filters is strongly affected by the sample weight[27]. The
IP T
addition of a sintered alumina coating also provided a remarkable improvement, as in a previous study by Emmel et al.[30] Dense alumina alone is much stronger than carbon-bonded alumina in
SC R
general, thanks to the presence of a ceramic bonding. Furthermore, another strengthening effect of
the Al2O3 coating is related to compressive residual stresses that are induced into the filter struts due to the difference of thermal expansion coefficients between carbon-bonded alumina substrate and
U
Al2O3 coating. Finally, the coated filters were also heavier than the uncoated substrates. In case of
N
batch AC5-2AH-alox, the CCS was increased to 0.25 ± 0.04 MPa. This means that the filters broke at a
A
force of about 625 N, which is in the range of technically applied filters. Concluding, the samples
M
produced with the new binder system showed a lower CCS when compared to the pitch-bonded filters. We could not confirm whether an addition of silicon significantly improves the strength of the
ED
material. However, the addition of a pure sintered alumina coating was clearly beneficial in this
PT
regard.
Table 3: Cold crushing strength of the filters (AC5 indicates the reference formulation with pitch
A
CC E
binder).
Batch
CCS [MPa]
Mass [g]
AC5
0.33 ± 0.07
20.01 ± 0.54
AC5-2AH
0.05 ± 0.01
16.94 ± 0.35
AC5-2AH-alox
0.25 ± 0.04
24.54 ± 0.47
AC5-2AHSi
0.09 ± 0.02
18.44 ± 0.31
AC5-2AHSi-alox
0.33 ± 0.09
24.72 ± 0.22
15
3.5 Oxidation behavior The results of the oxidation test on the filters prepared with the new binder system are presented in Figure 6. A sample produced with the pitch binder was also tested as a reference. This one showed the largest total mass loss, of about 35 %. However, an initial loss of 1 % below 100 °C was noticed only for this sample. This was likely related to water evaporation after prolonged storage, as
IP T
reported by Luchini et al. for the same kind of filters [37]. The total mass loss for AC5-2AH and AC5-
2AHSi samples was only 20 and 25 %, respectively. Since the environment-friendly binder system has
SC R
a much lower carbon yield than pitch, after pyrolysis the new filters contained less carbon than the reference one. The presence of n-Si in the AC5-2AHSi sample resulted in an improved carbon yield, hence the mass loss due to oxidation was also larger. Interestingly, the carbon fraction in the AC5-
U
2AH and AC5-2AHSi filters started burning at about 500 °C, while in the AC5 it started already at 400
N
°C. This can be easily explained by the absence of antioxidants in the latter. The mass loss as a
A
function of the temperature (i.e. the slope of the curve) was fairly constant for the newly developed
M
filters. On the other hand, the AC5 sample showed 3 distinct regions, which identify the burnout of
ED
carbon fractions from Carbores P, carbon black, and graphite, respectively [35]. At 800 °C, oxidation of
A
CC E
PT
all filters was already completed.
Fig. 6: Oxidation curves from thermogravimetric analysis up to 1000 °C in air.
16
3.6 XRD Figure 7 presents XRD patterns of the AC5-2AH and AC5-2AHSi samples, and of the coated (-alox) variations after removal of the alumina coating. The X-ray diffractograms of all formulations clearly showed the peaks of aluminum oxide in the alpha form. In addition, in all samples a pronounced graphite peak was detected at approx. 2θ = 26.5°, together with other minor peaks also from
IP T
graphite. The new formulations showed the same features as the AC5 reference[35]. No remarkable
difference was observed between the AC5-2AH and the AC5-2AHSi sample: obviously the Si addition
SC R
was too small to be detected by the instrument. The relative intensities of the graphite peaks were
also very similar, so the effect of n-Si addition could not be verified here. However, after sintering of the alumina coating (Tmax = 1400°C), the intensity of the main graphite peak was higher, likely as a
U
result of crystallization of part of the carbon. More detailed information was obtained via Raman
A
CC E
PT
ED
M
A
N
spectroscopy (see next section).
Fig. 7: X-ray diffraction patterns of the new formulations and of the coated filters 17
3.7 Raman spectroscopy For the Al2O3/Tannin/Lactose/Hexa/SiO2 samples at room temperature a very strong photoluminescence background was observed, so that no Raman spectra could be measured using 532 nm as excitation wavelength. Strong photoluminescence background was also measured using 442 nm and 633 nm laser lines as excitations. This photoluminescence background was probably
IP T
connected with the O-bridges in the polyphenol structure of the tannin molecules and the
localisation of excitons on the phenol rings. Strong photoluminescence background was also
[38]
SC R
observed in literature for tannin-furanic rigid foams and was attributed to the high hydrogen content . High photoluminescence covering the Raman spectra was found to be a typical signature of
hydrogenated carbon samples with high hydrogen content. This was most likely related to saturation
U
of non-radiative recombination centres[33]. The samples were then annealed at 200°C, 300°C, 400°C
N
and 500°C and the Raman spectra measured at room temperature using 532 nm as excitation (Fig.
A
8a). The spectra were dominated by the intense luminescence signal of Al2O3 at 4389.6 cm-1 (Fig. 8)
M
and 4359.8 cm-1 (Fig. 9) which correspond to electron transitions of Cr3+ in corundum[39][40]. In
ED
addition the D and G bands, characteristic for disordered carbon systems, were present (this spectral region is better seen as a zoom in figure 8b)[Rob02]. The G-band at ~ 1600 cm-1 corresponds to the
PT
in-plane bond stretching movement of pairs of sp2 C atoms (all sp2 sites, not only in aromatic rings), while the D-band at ~ 1350 cm-1 is a breathing mode connected to the presence of sixfold aromatic
CC E
rings being activated by disorder [32][34]. The spectrum of the sample annealed at 200°C still showed a high photoluminescence background,
A
which was however lower than the value measured before annealing. This could indicate that some of the O-bridges and of the -OH groups were broken at 200°C and smaller aromatic parts remained, where the exciton was still localised on the phenol rings. In the case of amorphous hydrogenated carbon films, it was shown that the photoluminescence comes from isolated sp2 carbon clusters (as rings) and its intensity is lower when the hydrogen content decreases[41]. Our system annealed at these temperatures was different from the a-C:H films but still a system containing disordered 18
carbon. As one can see in the inset of Fig. 8a the intensity of the photoluminescence at 4200 cm-1 decreased strongly with increasing annealing temperature. This indicates that the sp2 carbon ring clusters were growing (becoming less isolated) and the system started to lose hydrogen with increasing temperature. Most likely in the 300-400°C temperature range the neighbouring C-OH groups were condensed. This induced on the one hand a decrease of the O and H content in the
IP T
system, and on the other hand clustering of different sp2 carbon aromatic parts. In a similar temperature range, using infrared spectroscopy it was shown that neighbouring Si-OH groups are
SC R
forming Si-O-Si bonds and water[42]. Formation of larger sp2 carbon aromatic clusters by breaking of
C-O bonds and -OH condensation was the mechanism proposed for synthesizing tannin-based Si codoped carbon for supercapacitor applications by a microwave-assisted method[43]. For our sample
U
the spectra in the Fig 8b (right panel) show the region of the -OH vibrations at ~3400 cm-1 with a clear
N
decrease of intensity for the sample which was annealed at 500°C. The spectra also show a broad
A
Raman band at ~2900 cm-1 which is a superposition of the C-H stretching vibrations with the second
M
order of the C-C vibration and thus difficult to interpret.
ED
Now we focus on the Raman spectra in the region of C-C vibrations. In nano-graphite systems, the decrease of the ID/IG intensity ratio is attributed to improved ordering and growth of the
PT
nano-crystals dimension. In contrast, in the case of more disordered amorphous carbon systems with small clusters, the interpretation of the ID/IG ratio is exactly opposite[32]. Thus, the increase of the D-
CC E
band and consequently of the ID/IG ratio indicates an increase of the size of the sp2 carbon clusters indicating ordering in the amorphous carbon system. The Figure 8b (left panel) shows the Raman spectra as a zoom of the region of the D- and G-bands. It should be mentioned that the D and G band
A
region in the Raman spectra of tannin-furanic rigid foams were interpreted by Reyer et al. in terms of rearrangement of furanic and tannic structures during polymerisation[38]. As can be seen, increasing the temperature for the Al2O3/Tannin/Lactose/Hexa/SiO2 samples from 200°C to 500°C induced an increase of the ID/IG ratio. For the ID/IG ratio (peak area ratio) values of 0.16, 1,17, 1,34 and 1,91 were obtained, by fitting the D-band at ~1387 cm-1 and G-band at ~1593 cm-1 using Gauss functions, for 19
the temperatures 200°C, 300°C, 400°C and 500°C, respectively. This increase of the ID/IG ratio with increasing annealing temperature may be associated to a growth of the sp2 carbon cluster area. This happened most likely because part of the OH groups from phenols were released due to thermal energy and the smaller aromatic molecular fragments connected forming larger aromatic sp2-carbon clusters. This behaviour is in very good agreement with the temperature evolution of the
increase of size of the sp2 carbon clusters with increase of the temperature too.
SC R
b)
ED
M
A
N
U
a)
IP T
photoluminescence which was previously discussed and indicated less hydrogen in the system and an
PT
Fig. 8: Raman spectra measured at room temperature of the Al2O3/Tannin/Lactose/Hexa/SiO2 samples annealed at 200°C, 300°C, 400°C and 500°C. The inset shows the value of photoluminescence at 4200 cm-1 as
CC E
a function of annealing temperature a) The same spectra zoomed in the regions of the D- and G- C-C vibrational bands and in the region of -OH stretching band. b)
A
Figure 9 shows the spectra measured at room temperature for annealed samples of
Al2O3/Tannin/Lactose/Hexa/SiO2 to which n-doped Si was added. In all spectra the Si phonon mode at 520 cm-1 and the D and G bands corresponding to C-C vibrations can be observed. If we compare the spectra to those in Figure 8, one can see that the photoluminescence intensity was lower for the samples mixed with n-doped Si. We speculate that n-Si can act as a catalyser of the clustering processes mentioned before. On the other hand, the lower photoluminescence intensity may also be 20
related to lower carbon content in the measured volume (part of the excited volume is now occupied by Si). However also in these spectra we observed an increase of the ID/IG ratio with annealing temperature, that was found by Ferrari et al. [32] to be "proportional to the number and clustering of rings", and indicates an increase of ordering in the amorphous carbon system. In the case of diamond-like carbon films where Si was co-deposited, increasing Si content was found to decrease
IP T
the ID/IG ratio, which was connected with a decrease of the sp2 bonded clusters and promotion of sp3 formation [44]–[47]. In our case the situation was different because n-doped Si was added to the surface
PT
ED
M
A
N
U
SC R
of the samples, so that an atomic mixing was not expected.
CC E
Fig. 9: Raman spectra measured at room temperature of the Al2O3/Tannin/Lactose/Hexa/SiO2 samples mixed with n-doped Si annealed at 200°C, 300°C, 400°C and 500°C. In order to better visualize the influence of n-Si addition to our samples, we directly
A
compared spectra measured at room temperature for the samples annealed with and without n-Si. Figure 10a shows the spectra measured for samples which have been annealed at 500°C and the figure 10b for a sample coked at 800°C. In both cases one can see the second order peak of Si at ~ 960 cm-1. The spectra are shown scaled to the same intensity of the D and G peaks. The spectra of the samples mixed with n-Si appeared to be noisier due to the smaller Raman excited carbon content, because part of the excited volume was occupied by Si. It is very clear for both annealing 21
temperatures that the addition of n-Si caused a decrease of the tail at ~1160 cm-1 and an increase of the signal at ~ 1470 cm-1 (indicated by arrows). Besides the known D- and the G-bands, the region of the C-vibrations generally shows additional features (at 1100-1200 cm-1 and 1400-1500 cm-1) which are lower in intensity and therefore are either neglected or have controversial interpretation in literature. Exactly in the region
IP T
were the n-Si caused changes in our spectra Ferrari et al. [48] and Piazza et al. [49] assigned Raman
bands at ~1150 cm-1 and at ~1460 cm-1 to transpolyacetilene segments. The band at lower energy
SC R
corresponds to a C-C vibration, while the band at higher energy to a C=C vibration in sp2 C chains in
transpolyacetilene [50]. In our complex carbon system, besides C rings also C chains may be present. The addition of n-Si to one sample clearly modified the carbon bonding. These two bands were found
U
to exhibit a strong dispersion with changing excitation energy [48]. Measuring Raman spectra with
N
different laser energy in order to check the dispersion of these modes could bring more information
M
A
regarding the C bonding in our system, but is beyond the aim of this paper. b)
A
CC E
PT
ED
a)
Fig. 10: Comparison of Raman spectra measured at room temperature for samples with and without n-Si annealed at 500°C (a) and coked at 800°C (b).
22
3.8 Impingement tests Fig. 11 shows the filters in the sand molds after the impingement tests. For every filter batch, three tests were carried out. All samples of AC5-2AH were destroyed when the molten steel was poured through the filters, #1. This result was expected, due to the very low CCS value of 0.05 MPa. The other three batches (AC5-2AHSi; AC5-2AH-alox; AC5-2AHSi-alox) survived the thermal shock and the
IP T
impingement itself. It was shown above that the addition of n-Si increased the CCS value. All AC5-
2AHSi samples were also intact after the experiment, #2 in Fig. 11. It is obvious that the Al2O3 coating
SC R
had a beneficial effect due to the higher CCS values. Therefore, the impingement test was positive
A
CC E
PT
ED
M
A
N
U
and the filters were also intact, #3 and #4 in Fig. 11.
Fig. 11: Foams in sand mold after impingement tests: 1=AC5-2AH; 2=AC5-2AHSi; 3=AC5-2AH-alox; 4=AC5-2AHSi-alox
23
4. Conclusions In the present work a new development of carbon-bonded alumina filters for steel melt filtration was presented. Samples were manufactured by replacing the conventional pitch binder Carbores®P, with a simple and environmental friendly system based on lactose and tannin. These two, unlike conventional binders derived from oil refining, are absolutely harmless and renewable. The lactose-
IP T
tannin binder system forms a resit network structure comparable to novolac resins. The lower
carbon yield compared to pitches resulted in filters with thinner struts and more cracks after firing.
SC R
This defects led to a relatively low cold crushing strength in comparison to the reference filters. The addition of n-Si increased the strength of the filters by improving the carbon yield of the binder system. An additional Al2O3 coating which sintered during heat treatment increased the CCS values
U
significantly. Raman spectroscopic measurements carried out at room temperature for
N
Al2O3/Tannin/Lactose/Hexa/SiO2 samples annealed at 200°C, 300°C, 400°C and 500°C showed a
A
reduced photoluminescence background and the ID/IG ratio increased with annealing temperature.
M
This behaviour may be explained by -OH thermal release and the formation of larger aromatic carbon
ED
ring clusters. The new filters were successfully tested by impingement with molten steel without
PT
preheating.
CC E
Acknowledgements
The authors would like to thank Ms. E. Qoku for X-ray diffraction measurements and Dr. G. Schmidt
A
for the SEM investigations. We also thank the German Research Foundation (DFG) for supporting these investigations within the Collaborative Research Centre 920 “Multi-Functional Filters for Metal Melt Filtration – A Contribution towards Zero Defect Materials”, subprojects A01, A04 and S03.
24
REFERENCES D. Janke, K. Raiber, Grundlegende Untersuchungen Zur Optimierung Der Filtration von Stahlschmelzen, 1996.
[2]
C.G. Aneziris, D. Borzov, J. Ulbricht, J. Suren, Cfi-Bericht DKG 2003, 80, E31.
[3]
C.G. Aneziris, D. Borzov, J. Ulbricht, J. Suren, in Unitecr, Osaka 2003, 631.
[4]
C.G. Aneziris, D. Borzov, J. Ulbricht, Interceramic Refract. Man. 2003, 22.
[5]
D. Borzov, J. Ulbricht, W. Schulle, in Int. Feuerfest-Kolloquium Aachen, 2001, 50.
[6]
Verordnung (EG) Nr. 1907/2006 (REACH): Registration, Evaluation, Authorisation and Restriction of Chemicals, 2006.
[7]
C. Biermann, Entwicklung Eines Neuen Umweltfreundlichen Bindemittel-Systems Für Die Feuerfestindustrie Auf Pharmazeutischer Und Lebensmittlechemicher Basis, TU Bergakademie Freiberg, 2016.
[8]
E. Gueguen, C.G. Aneziris, C. Biermann, Feuerfeste Formkörper Und Massen Sowie Bindemittel Und Verfahren Zu Deren Herstellung, 2017, WO 2017118449 A1.
[9]
M. Hampel, C.G. Aneziris, Cfi-Bericht DKG 2007, 84, E125.
[10]
M. Hampel, C.G. Aneziris, Tech. Keramische Werkstoffe 2007.
[11]
A. Gardziella, J. Suren, Stahl Und Eisen 1999, 119, 135.
[12]
A. Gardziella, J. Suren, Stahl Und Eisen 1993, 113, 75.
[13]
C.G. Aneziris, J. Hubálková, R. Barabás, J. Eur. Ceram. Soc. 2007, 27, 73.
[14]
U. Klippel, Interceramic Refract. Man. 2006, 1, 14.
[15]
U. Klippel, in 49. Int. Feuerfest-Kolloquium, Aachen 2006, 6.
[16]
U. Klippel, V. Stein, C.G. Aneziris, in UNITECR ’07, 2007.
[17]
V. Stein, C.G. Aneziris, U. Klippel, W. Schönwelski, E. Guéguen, in Unitecr, Salvador 2009.
[18]
S. Zhang, N.J. Marriott, W.E. Lee, J. Eur. Ceram. Soc. 2001, 21, 1037.
[19]
S. Zhang, W.E. Lee, J. Eur. Ceram. Soc. 2001, 21, 2393.
[20]
W.E. Lee, S. Zhang, Int. Mater. Rev. 1999, 44, 77.
[21]
A. Yamakuchi, Taikabutsu Overseas 1999, 4, 14.
[22]
A. Watanabe, H. Takahashi, S. Taganaga, Taikabutsu Overseas 1987, 4, 17.
A
CC E
PT
ED
M
A
N
U
SC R
IP T
[1]
[23]
A. Yamaguchi, S. Zhang, J. Yu, J. Am. Ceram. Soc. 1996, 79, 2509.
[24]
V. Stein, Contribution to the Characteristic Improvement of Carbon Bonded Doloma Refractories by Addition of Functional Ceramic Materials, TU Bergakademie Freiberg, 2011.
[25]
V. Stein, C.G. Aneziris, J. Ceram. Sci. Technol. 2014, 5, 115.
[26]
J.E. Lang, A. Karl, H. Rauleder, E. Müh, G. Stochniol, Verfahren Zur Pyrolyse von Kohlenhydraten, 2010, WO 2010/037699 A2.
[27]
M. Emmel, C.G. Aneziris, Ceram. Int. 2012, 38, 5165. 25
J. Luyten, S. Mullens, J. Cooymans, A.M. De Wilde, I. Thijs, R. Kemps, J. Eur. Ceram. Soc. 2009, 29, 829.
[29]
K. Schwartzwalder, A. V Somers, Method of Making Porous Ceramic Articles, 1963, 3,090,094.
[30]
M. Emmel, C.G. Aneziris, J. Mater. Res. 2013, 28, 2234.
[31]
M.A. Tamor, W.C. Vassell, J. Appl. Phys. 1994, 76, 3823.
[32]
A.C. Ferrari, J. Robertson, Phys. Rev. B 2000, 61, 14 295.
[33]
C. Casiraghi, A.C. Ferrari, J. Robertson, Phys. Rev. B - Condens. Matter Mater. Phys. 2005, 72, 1.
[34]
G. Irmer, A. Dorner-Reisel, Adv. Eng. Mater. 2005, 7, 694.
[35]
M. Emmel, Development of Active and Reactive Carbon Bonded Filter Materials for Steel Melt Filtration, TU Freiberg, 2014.
[36]
A. Weidner, D. Krewerth, B. Witschel, M. Emmel, A. Schmidt, J. Gleinig, O. Volkova, C.G. Aneziris, H. Biermann, Steel Res. Int. 2016, 87, 1038.
[37]
B. Luchini, J. Grabenhorst, J. Fruhstorfer, V.C. Pandolfelli, C.G. Aneziris, J. Am. Ceram. Soc. 2018, 1.
[38]
A. Reyer, G. Tondi, R.J.F. Berger, A. Petutschnigg, M. Musso, Vib. Spectrosc. 2016, 84, 58.
[39]
T.H. Huang, C.C. Hsu, C.T. Kuo, P. Lu, W.S. Tse, D.P. Wang, T.C. Chou, A.Y.G. Fuh, J. Appl. Phys. 1994, 75, 3599.
[40]
C. Röder, T. Weißbach, C. Himcinschi, J. Kortus, S. Dudczig, C.G. Aneziris, J. Raman Spectrosc. 2014, 45, 128.
[41]
B. Marchon, Jing Gui, K. Grannen, G.C. Rauch, J.W. Ager, S.R.P. Silva, J. Robertson, IEEE Trans. Magn. 1997, 33, 3148.
[42]
C. Himcinschi, M. Friedrich, K. Hiller, T. Gessner, D.R.T. Zahn, Semicond. Sci. Technol. 2004, 19, 579.
[43]
S.K. Ramasahayam, U.B. Nasini, A.U. Shaikh, T. Viswanathan, J. Power Sources 2015, 275, 835.
[44]
J.F. Zhao, P. Lemoine, Z.H. Liu, J.P. Quinn, J.A. McLaughlin, J. Phys. Condens. Matter 2000, 12, 9201.
[45]
P. Papakonstantinou, J.. Zhao, P. Lemoine, E.T. McAdams, J.A. McLaughlin, Diam. Relat. Mater. 2002, 11, 1074.
[46]
G.A. Abbas, J.A. McLaughlin, E. Harkin-Jones, Diam. Relat. Mater. 2004, 13, 1342.
[47]
K. Er, M. So, J. Ceram. Process. Res. 2011, 12, 187.
A
CC E
PT
ED
M
A
N
U
SC R
IP T
[28]
[48]
A.C. Ferrari, J. Robertson, Phys. Rev. B 2001, 63, 121405.
[49]
F. Piazza, A. Golanski, S. Schulze, G. Relihan, Appl. Phys. Lett. 2003, 82, 358.
[50]
E. Mullazzi, G.P. Brivio, E. Faulques, S. Lefrant, Solid State Commun. 1983, 46, 851.
26
S0955-2219(18)30526-0 https://doi.org/10.1016/j.jeurceramsoc.2018.08.029 JECS 12054
To appear in:
Journal of the European Ceramic Society
Received date: Revised date: Accepted date:
14-5-2018 14-8-2018 20-8-2018
Please cite this article as: Himcinschi C, Biermann C, Storti E, Dietrich B, Wolf G, Kortus J, Aneziris CG, Innovative carbon-bonded filters based on a new environmentalfriendly binder system for steel melt filtration, Journal of the European Ceramic Society (2018), https://doi.org/10.1016/j.jeurceramsoc.2018.08.029 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Innovative carbon-bonded filters based on a new environmental-friendly binder system for steel melt filtration
a
IP T
C. Himcinschia,*, C. Biermannb, E. Stortib, B. Dietrichc, G. Wolfc, J. Kortusa, C.G. Anezirisb
Institute of Theoretical Physics, TU Bergakademie Freiberg, Leipziger Straße 23, 09599 Freiberg,
b
SC R
Germany
Institute of Ceramic, Glass and Construction Materials, TU Bergakademie Freiberg, Agricolastraße
17, 09599 Freiberg, Germany c
U
Foundry Institute, TU Bergakademie Freiberg, Bernhard-von-Cotta-Straße 4, 09599 Freiberg,
M
ED
* [email protected]
A
N
Germany
PT
Abstract:
New carbon-bonded alumina filters for steel melt filtration were developed. The carbonaceous
CC E
matrix was based on a new, environmental friendly binder system based on lactose and tannin. The filter preparation was analogous to the production of conventional foam filters according to the Schwartzwalder process. The processing as well as the rheology of the slurries was investigated. An
A
addition of n-Si increased the carbon yield and the cold crushing strength (CCS) of the samples. Higher values of CCS were obtained after coating of the filters with alumina. The material was characterized by scanning electron microscopy, X-ray diffraction and Raman spectroscopy. The applicability of these new filters was assessed in impingement tests with a steel melt, in which three out of four recipes survived the thermal shock. 1
Keywords: Steel filtration; Lactose; Tannin; Ceramic foam filters; Raman Spectroscopy
1. Introduction
IP T
Carbon-bonded ceramic foam filters are established for applications in steel melt filtration, to
remove non-metallic inclusions from the molten steel at temperatures between 1400 and 1650°C,
SC R
and to guarantee the reliability required by the customer, especially for safety-relevant components[1].
U
The state of art is the use of novolac or resoles as binder[2]–[5]. Pitch binders have the disadvantage of
N
releasing harmful components such as carcinogenic products (benzo[a]pyrene) during production
A
and operation. An improvement was achieved by the application of Carbores®P (CARBOnaceous
M
RESin), which is a kind of high melting artificial coal-tar resin. The yield of benzo[a]pyrene for Carbores is much lower than for traditional pitches, 50000 vs < 300 ppm. However, the limit in
ED
Germany is fixed at 50 ppm[6]. Thus, environmental friendly resins are generally preferred. Recently, a new environmental friendly alternative binder system for carbon-bonded MgO-C materials was
PT
presented[7]. In addition, a patent describing it was applied[8]. The structure of this new binder
CC E
system, based on lactose and tannin, is similar to the novolac structure. During the curing step, a resit network structure comparable to the novolac lattice is formed. The stability of the cured state is also similar[7]. In the residual carbon, the structure of the resit network is an isotropic glassy phase.
A
This isotropic structure leads to a lower oxidation and corrosion resistance and worse thermomechanical properties compared to more ordered carbon forms such as graphite[9][10][11][12]. A way to enhance the quality of carbon-bonded refractory products is the addition of antioxidants[13], which hinder carbon oxidation. During processing, metallic antioxidants, e.g. Al are added to the raw materials[13]–[19]. This addition results in the improvement of oxidation resistance and of the mechanical strength[18][20]–[22]. 2
Yamaguchi et al. [23]determined the effect of certain oxides on the oxidation resistance of graphite and amorphous carbon. For instance, TiO2 prevents the oxidation of graphite by donating electrons and therefore stabilizing its structure. The new environmental friendly binder system is a combination of several raw materials and additives[7]: hexa for hardening similar to novolac binders, n-Si for increasing the carbon yield[24][25],
IP T
TiO2 as electron donator [13][24] and Al as antioxidant [13]. A mixture of lactose and tannin showed a large volume expansion during heat treatment and the resulting product was extremely porous.
SC R
Thus, SiO2 was added as antifoam agent [26]. The average carbon yield of the new binder system after treatment in reducing atmosphere at a temperature of 1000 °C was 51.5 ± 0.5 %, which is similar to
U
the yields of the conventional novolac binders[7].
N
Open-cells foams are the most widely used structures in the case of metal melt filtration[27].
A
However, so far only bulk MgO-bricks were produced on basis of the new binder system.
techniques was given by Luyten et al.[28].
M
Several methods for production of foam structures were developed. An overview of the different
ED
In the present work, ceramic foams based on the new environmental friendly binder system were manufactured according to the replica technique patented by Schwartzwalder in 1963[29].
PT
Polyurethane foams were used as base for the replica technique. The porosity (defined as pores per
CC E
inch, ppi) of foam filters for metal filtration is always chosen based on the metal properties. For example, foams with 10 ppi and 30 ppi are commonly used for steel and for aluminum melt filtration, respectively[27]. Thus, in our case, 10 ppi foams were used. After the impregnation step, a thermal
A
treatment was applied in order to cure the binder and decompose the polyurethane foam. After treatment, the material consists of a carbonaceous matrix surrounding the ceramic particles. The foams possess an open-cell structure with cells of pentagon-dodecahedronal geometry linked by triconcave struts[27]. Moreover, active oxide coatings were applied to some of the filters in order to increase the cold crushing strength and to improve the collection of oxide particles from the steel
3
melt during operation[30]. The foam filters were characterized in terms of structure and mechanical properties and tested in impingement tests in contact with steel melt. The structure of the lactose-tannin-based binder system and in particular the effect of n-Si was studied by Raman spectroscopy. Raman spectroscopy is a very sensitive detection method for specific vibrational fingerprints of distinct atomic groups in molecules, crystalline or amorphous
IP T
systems. Thus, it is suitable for the identification of the chemical structure and / or changes in the chemical structure. Nowadays, Raman spectroscopy is widely recognized as a non-destructive
SC R
standard method for the characterization of carbon-containing materials because of the "fingerprint"
U
and information on bonding that it can deliver for such systems [31]–[34].
N
2. Materials and characterization techniques
A
The raw materials used for the preparation of the investigated Al2O3-C foams were aluminum oxide
M
(Martoxid MR70, Martinswerk, Germany, 99.80 wt.% Al2O3, d90 ≤ 3.0 µm), lactose (α-Lactose-
ED
monohydrat DAB 9, DM-Markt, Germany) and tannin (Quebrachoextrakt, type Indusol ATO., Otto Dille, Germany) used both as binders and carbon sources[7], hexamethylenetetramine (Momentive,
PT
Germany), fine natural graphite (AF 96/97, Graphit Kropfmühl, Germany, 96.7 wt.% carbon, 99.8 wt.% < 40 µm), and carbon black powder (Luvomaxx N-991, Lehmann & Voss &Co., Germany, ≥99.0
CC E
wt.% carbon, >0.01 wt.% ash content, primary particle size: 200-500 nm). The following additives were used: ammonium-ligninsulfonate C12C (Otto Dille, Germany) as temporary binder and wetting agent, Castament VP 95 L (BASF, Germany) as dispersing agent, Contraspum K 1012 (Zschimmer &
A
Schwarz, Germany) as antifoam agent[35]; TiO2 (nano-powder TR, ≥ 99 wt.% rutile, Crenox GmbH, Germany) and aluminum metal (Al-Grieß – fraction 0-75 µm, ≥ 99.6 % Al, d50 = 29.0 µm, Hoesch Granules GmbH, Germany) as antixoxidant agents and SiO2 (RW-Filler weiß, RW silicium GmbH, Germany) as antifoam agent[7]. To increase the carbon yield, P-doped n-Si (grade 1a, > 99.9 % Si, particle size < 63 µm, Silchem, Germany) was used in one batch. 4
The preparation of the carbon-bonded filters was divided into two steps according to the procedure described by Emmel et al.[27]. In the first step, a highly viscous slurry for impregnation of the PU foam was prepared in a Hobbart-type mixer (ToniTechnik, Germany). Here, the powder components listed in Table 1 were weighted and dry mixed for 5 min. Then, the liquid additives (Castament VP 95 L, Contraspum K 1012) were dispersed in deionized water and added to the dry mixture. Further
IP T
deionized water was added step by step until a plastic mass was obtained. This mass was mixed for 5 min for best homogenization. Finally, the rest of deionized water was added to get an impregnating
SC R
slurry with the total solid content of 78 wt.%.
Table 1: Composition of the impregnating and spraying slurries without (recipe 1: AC5-2AH) and with
U
n-Si (recipe 2: AC5-2AHSi), and of the alumina coating slurry.
Coating material
N
Filter material Weight fraction (%)
Al2O3 Martoxid MR 70
66.0
lactose/tannin=5/1
20.0
-
Carbon black N-991
6.3
-
7.7
-
M
ED
100
Mass in wt.% (related to total solid content)
Hexametilenetetramine
10.0 b
-
TiO2
0.5
-
Al
0.1
-
A
CC E
Additives
PT
Graphite AF 96/97
A
Material
SiO2
4.0
-
n-Si (recipe 2)
5.0 b
-
Castament VP 95 L
0.3
0.3
Contraspum K 1012
0.1
-
5
Ammonium-ligninsulfonate C12C
1.5
1.5
Total solid content (wt.%) Impregnating slurry
78.0
-
Spraying slurry
70.0
65.0
b
related only to binder content of 20 wt.%
IP T
This slurry was used for the impregnation of PU-foam blocks with dimensions of 50 mm x 50 mm x 22 mm. For this step the slurry was pressed by hand into the foams to guarantee a complete coating of
SC R
the PU framework. To remove the excess slurry, the foam blocks were then pressed through a special roll-pressing device with two counter-rotating rolls. The gap between the two rolls was in our case 4
U
mm. Subsequently, the wet impregnated foams were dried at room temperature to a constant
N
weight. A further processing step was necessary to obtain foams with a homogenous and sufficient
A
thickness of ceramic layer. Thus, a spraying slurry with a total solid content of 70 wt.% was prepared as described above. This slurry was applied by spraying by means of an airgun HS-25HVLP
M
(Krautzberger, Germany) at a spraying pressure of 3 bar in a spraying chamber. The distance between
ED
the spray gun (nozzle) and the foam was kept at approximately 10 cm. The four smaller sides were sprayed for about 2-3 s and the two larger sides for 5-6 s. The weight of each foam at the end of the
PT
spraying step was about 27 g in the wet condition. Afterwards, the samples were dried at room
CC E
temperature for 24 h.
The dried foams were hardened inside a drying chamber up to a temperature of 180°C according to the schedule in Fig. 1. This treatment was followed by pyrolysis in a retort filled with calcined petcoke
A
(Müco, Germany; particle size 0.2 – 2 mm), in order to approach reducing conditions. The heating rate was 1 °C/min up to 1000°C and 180 min holding time. In addition, every 100 °C a holding step of 30 min was included to prevent damage from volatile release. After the pyrolysis process the samples cooled down to room temperature inside the retort in the oven. The final weight of the carbonbonded filters was approximately 18 g.
6
b)
SC R
IP T
a)
Fig.1: Heating procedure for hardening (a), pyrolysis of the filters (b, blue line) and sintering of the
U
extra coating (b, orange line)
N
Furthermore, some coked samples were spray coated with a functional coating based on pure
A
alumina[30][36]. The composition of this slurry is presented in Table 1. The coating procedure was
M
similar to the previous one, described above. In this case, the weight of the wet filters reached about 29 g. The second coating was also dried at room temperature. Afterwards, the samples were treated
ED
again inside a retort filled with coke. The heating rate was again 1 °C/min up to 300 °C with 30 min
of the coating.
PT
holding time; 3 °C/min up to 1400 °C and 5 h dwell time, to promote sintering of the alumina grains
CC E
In the end four different recipes were produced, i.e. (i) without n-Si and Al2O3 coating (AC5-2AH) and (ii) with Al2O3 coating (AC5-2AH-alox) as well as (iii) with n-Si without Al2O3 coating (AC5-2AHSi) and (iv) with Al2O3 coating (AC5-2AHSi-alox), respectively, see Fig. 2.
A
The rheological behavior of the slurries was measured with the aid of a rheometer RheoStress RS 150 (Haake, Germany). For the measurement of the flow curves, coaxial cylinder measuring systems of the type Z38 and Z40 DIN were used for the impregnating and spraying slurries, respectively. The flow behavior was determined by measuring the viscosity and shear stress under variation of the
7
shear rate. This was changed linearly from 0.1 to 500 s-1 (up to 1000 for the spraying slurries), and
PT
ED
M
A
N
U
SC R
IP T
kept at this level for 90 s, before decreasing it at the same rate down to the initial 0.1 s-1 value.
CC E
Fig. 2: Carbon-bonded Al2O3-C filters without (A, B) and with (C, D) n-Si; without Al2O3 coating (A, C) and with Al2O3 coating (B, D)
A
The microstructural characterization of the filters after heat treatment was carried out by means of a digital light microscope (VHX-200 D, Keyence, Japan) and a scanning electron microscope, equipped with an EDS-detector (SEM XL30, Philips, Netherlands). In order to determine the thickness of the sprayed coating, broken struts were investigated in cross section. The open porosity was evaluated using a mercury intrusion porosimeter AutoPore V 9600 from Micromeritics Instrument Corporation (USA) and the software Micro Active. 8
The X-ray diffraction analysis was carried out on a standard diffractometer (PANalytical X’Pert Pro MPD 3040/60, Almelo, Netherlands). The instrument operates in Bragg-Brentano geometry with a fixed divergence slit and rotating sample stage. Standard Cu Kα radiation and a scan step size of 0.013° with 30 s holding time per step were used. The X-ray source was operated at 40 kV and 40 mA. The analyses were performed on ground filter struts using the back loading technique to
IP T
minimize preferred orientation. The cold crushing strength (CCS) was determined on a universal testing machine (TT 2420, TIRA
SC R
GmbH, Germany) using a 1 kN load cell. The displacement rate was 3 mm/min up to a force of 5 N. At this load the loading rate was changed to 1 MPa/s. The experiment was terminated when a strength loss of 80 % was reached. From each batch, five specimens were tested in order to obtain average
N
U
values and standard deviations.
A
The specimens for the Raman spectroscopy measurements were produced from a slurry without any
M
graphite and carbon black as well as without Al, TiO2 and n-Si. This slurry was cast into 30 mm x 30 mm square molds and dried at room temperature. Afterwards, two other slurries (one with TiO2 and
ED
a second with n-Si) were prepared. These slurries were cast in a thin layer on top of the dried previously cast substrates. A second drying step was also needed. Thus, several different specimens
PT
were prepared, i.e.:
CC E
(i) Al2O3/tannin/lactose/hexa/SiO2: with different thermal treatments (curing at 180 °C and coking at 200 °C, 300 °C, 400 °C and 500 °C) and (ii) Al2O3/tannin/lactose/hexa/SiO2/n-Si: with different thermal treatments (curing at 180 °C and
A
coking at 200 °C, 300 °C, 400 °C and 500 °C), respectively. The Raman measurements were performed using a Horiba Jobin Yvon Labram spectrometer equipped with a 600 mm-1 grating and a Peltier cooled CCD detector. The measurements were done in backscattering geometry. For excitation the 532 nm line of a frequency-doubled Nd-YAG laser was employed. In order to avoid the heating of the samples by the laser, a low laser power of 0.4 mW 9
was used. The light was guided and focused through a microscope objective with 20X magnification (ca. 10 µm focus diameter), which ensured a lower power density in comparison to microscope objectives with higher magnification and lower focus diameters. This ensured a negligible influence of the laser heating, except for the samples annealed at 200°C, for which a colour change was visible on the measured spot.
IP T
Finally, for the impingement tests, three filters of each batch were embedded in furan-bonded sand molds. Approximately 9 kg of steel were molten and directly poured on the samples, without
SC R
preheating them. The steel was always freshly molten before each filter impingement.
U
3. Results and discussion
A
N
3.1. Rheological behavior
M
In Figs. 3a and 3b the dynamic viscosity of the impregnating and spraying slurries as a function of the shear rate are shown. For comparison, the curves from the reference system (with pitch-based
ED
binder) are also included. A clear shear thinning behavior was observed for all slurries. As expected, the impregnating slurries exhibited a much higher dynamic viscosity than the spraying counterparts,
PT
as is it required for a proper impregnation process. The measurement for the AC5 reference had to be interrupted at 350 s-1 because of the limited shear achievable by the instrument. The same limit
CC E
was not reached by our slips since the reference impregnating slurry had a higher solid content (about 81.6 % versus 78 %). The rheological behavior at the investigated shear rates was very similar
A
to the reference, with dynamic viscosity around 3000 mPa·s at 300 s-1. The spraying slurries were ideal for the coating step, with a dynamic viscosity of about 100 mPa·s at 1000 s-1. Such behavior resulted in a very good sprayability, and at the same time in a good adhesion to the PU foam skeleton after the impact. Overall, no remarkable difference was observed when comparing the formulations without (AC5-2AH) and with n-Si (AC5-2AHSi), with the AC5-2AH sample showing only slightly higher viscosity. Despite having similar solids content, the reference spraying slurry was less 10
viscous than both samples prepared with tannin and lactose. This effect was likely due to the powder additives present in the new formulations, silica in particular.
N
U
SC R
IP T
a)
A
CC E
PT
ED
M
A
b)
Fig. 3: Rheology of slurries AC5-2AH and AC5-2AHSi for impregnation (a) and spraying (b), respectively.
11
3.2 Microstructure Optical micrographs of the carbon-bonded samples (without coating) are presented in Fig.4. The reference filter produced with the pitch binder showed thicker struts than the AC5-2AH and AC52AHSi. In addition, a large number of cracks were observed in particular on the AC5-2AH sample, which explain the lower cold crushing strength of this batch (see below). The addition of silicon
IP T
increased the carbon yield of the binder system, possibly limiting the formation of cracks in the AC52AH samples. Nevertheless, in order to obtain foams with structure similar to the reference one (i.e.
SC R
almost crack-free), it may be beneficial to extend the spraying process. Since the environmental
friendly binder system only yields 51.5 wt.% carbon against 80 wt.% of Carbores P, the filters will be sprayed to a slightly larger weight (i.e. 2-3 g in the wet condition) in future tests. Light blue particles
U
in the macropores of the AC5-2AHSi sample are coke residues, which probably interacted with some
A
N
silicon from the filter during the heat treatment at 1400 °C (metallic Si melts at 1414 °C).
M
In SEM investigations, broken struts were characterized. Fig. 5 shows characteristic struts of the four batches. In the center of the struts, large cavities left from the degradation of the initial PU foams
ED
with the typical triangular shape were present. There was no visible boundary between the material from the impregnating and from the spraying slurry. Large pores within the Al2O3-C filter struts were
PT
likely generated during heat treatment.
CC E
In Figs. 5b and 5d the Al2O3 coating is clearly visible as a homogenous and dense layer of material over the porous substrate. In fact, only the coating material was able to sinter during the heating treatment at 1400 °C, while the presence of carbon in the base material hindered this process. The
A
thickness of the alumina coating layer varied from about 50 µm to 100 µm.
12
IP T M
A
N
U
SC R
a)
A
CC E
PT
ED
b)
c) Fig. 4: Optical micrographs of (a) the reference filter (AC5), (b) AC5-2AH filter and (c) AC5-2AHSi, respectively.
13
b)
c)
d)
U
SC R
IP T
a)
A
M
coating (a, c) and with Al2O3 coating (b, d)
N
Fig. 5: Microstructure of the coked Al2O3-C filters without (a, b) and with (c, d) n-Si; without Al2O3
3.3 Open porosity
ED
The binder content of the new environmental friendly binder system was comparable to the conventional foam filters at 20 wt.%. However, due to the different evolution of volatiles during
PT
pyrolysis, we expected slightly different final porosities. The open porosity of the uncoated samples
CC E
of the lactose-tannin binder system was 50.8 % for AC5-2AH and 54.3 % for AC5-2AHSi. The conventional filters with 20 wt.% Carbores®P usually show an open porosity of about 30 %[27][35]. Thus, the new binder system yielded a much higher open porosity compared to traditional pitch
A
binder. Nevertheless, the filters could be easily handled after the heat treatment. 3.4 Cold crushing strength Table 3 gives the mean values and the standard deviations of the CCS of the four batches of filters and of the reference (AC5, bonded with Carbores P pitch). The CCS of the uncoated, Si-free batch was the lowest at 0.05 MPa, which was still enough to manipulate the filters but much lower than for the 14
reference samples. The use of n-Si as additive in the new binder system showed a positive influence on the cold crushing strength, which was nearly doubled. It should be pointed out, however, that these filters were slightly heavier than the ones without silicon, and that both batches prepared with the new binder system showed a lower weight than the reference samples. As reported by Emmel et al., the crushing strength of ceramic foam filters is strongly affected by the sample weight[27]. The
IP T
addition of a sintered alumina coating also provided a remarkable improvement, as in a previous study by Emmel et al.[30] Dense alumina alone is much stronger than carbon-bonded alumina in
SC R
general, thanks to the presence of a ceramic bonding. Furthermore, another strengthening effect of
the Al2O3 coating is related to compressive residual stresses that are induced into the filter struts due to the difference of thermal expansion coefficients between carbon-bonded alumina substrate and
U
Al2O3 coating. Finally, the coated filters were also heavier than the uncoated substrates. In case of
N
batch AC5-2AH-alox, the CCS was increased to 0.25 ± 0.04 MPa. This means that the filters broke at a
A
force of about 625 N, which is in the range of technically applied filters. Concluding, the samples
M
produced with the new binder system showed a lower CCS when compared to the pitch-bonded filters. We could not confirm whether an addition of silicon significantly improves the strength of the
ED
material. However, the addition of a pure sintered alumina coating was clearly beneficial in this
PT
regard.
Table 3: Cold crushing strength of the filters (AC5 indicates the reference formulation with pitch
A
CC E
binder).
Batch
CCS [MPa]
Mass [g]
AC5
0.33 ± 0.07
20.01 ± 0.54
AC5-2AH
0.05 ± 0.01
16.94 ± 0.35
AC5-2AH-alox
0.25 ± 0.04
24.54 ± 0.47
AC5-2AHSi
0.09 ± 0.02
18.44 ± 0.31
AC5-2AHSi-alox
0.33 ± 0.09
24.72 ± 0.22
15
3.5 Oxidation behavior The results of the oxidation test on the filters prepared with the new binder system are presented in Figure 6. A sample produced with the pitch binder was also tested as a reference. This one showed the largest total mass loss, of about 35 %. However, an initial loss of 1 % below 100 °C was noticed only for this sample. This was likely related to water evaporation after prolonged storage, as
IP T
reported by Luchini et al. for the same kind of filters [37]. The total mass loss for AC5-2AH and AC5-
2AHSi samples was only 20 and 25 %, respectively. Since the environment-friendly binder system has
SC R
a much lower carbon yield than pitch, after pyrolysis the new filters contained less carbon than the reference one. The presence of n-Si in the AC5-2AHSi sample resulted in an improved carbon yield, hence the mass loss due to oxidation was also larger. Interestingly, the carbon fraction in the AC5-
U
2AH and AC5-2AHSi filters started burning at about 500 °C, while in the AC5 it started already at 400
N
°C. This can be easily explained by the absence of antioxidants in the latter. The mass loss as a
A
function of the temperature (i.e. the slope of the curve) was fairly constant for the newly developed
M
filters. On the other hand, the AC5 sample showed 3 distinct regions, which identify the burnout of
ED
carbon fractions from Carbores P, carbon black, and graphite, respectively [35]. At 800 °C, oxidation of
A
CC E
PT
all filters was already completed.
Fig. 6: Oxidation curves from thermogravimetric analysis up to 1000 °C in air.
16
3.6 XRD Figure 7 presents XRD patterns of the AC5-2AH and AC5-2AHSi samples, and of the coated (-alox) variations after removal of the alumina coating. The X-ray diffractograms of all formulations clearly showed the peaks of aluminum oxide in the alpha form. In addition, in all samples a pronounced graphite peak was detected at approx. 2θ = 26.5°, together with other minor peaks also from
IP T
graphite. The new formulations showed the same features as the AC5 reference[35]. No remarkable
difference was observed between the AC5-2AH and the AC5-2AHSi sample: obviously the Si addition
SC R
was too small to be detected by the instrument. The relative intensities of the graphite peaks were
also very similar, so the effect of n-Si addition could not be verified here. However, after sintering of the alumina coating (Tmax = 1400°C), the intensity of the main graphite peak was higher, likely as a
U
result of crystallization of part of the carbon. More detailed information was obtained via Raman
A
CC E
PT
ED
M
A
N
spectroscopy (see next section).
Fig. 7: X-ray diffraction patterns of the new formulations and of the coated filters 17
3.7 Raman spectroscopy For the Al2O3/Tannin/Lactose/Hexa/SiO2 samples at room temperature a very strong photoluminescence background was observed, so that no Raman spectra could be measured using 532 nm as excitation wavelength. Strong photoluminescence background was also measured using 442 nm and 633 nm laser lines as excitations. This photoluminescence background was probably
IP T
connected with the O-bridges in the polyphenol structure of the tannin molecules and the
localisation of excitons on the phenol rings. Strong photoluminescence background was also
[38]
SC R
observed in literature for tannin-furanic rigid foams and was attributed to the high hydrogen content . High photoluminescence covering the Raman spectra was found to be a typical signature of
hydrogenated carbon samples with high hydrogen content. This was most likely related to saturation
U
of non-radiative recombination centres[33]. The samples were then annealed at 200°C, 300°C, 400°C
N
and 500°C and the Raman spectra measured at room temperature using 532 nm as excitation (Fig.
A
8a). The spectra were dominated by the intense luminescence signal of Al2O3 at 4389.6 cm-1 (Fig. 8)
M
and 4359.8 cm-1 (Fig. 9) which correspond to electron transitions of Cr3+ in corundum[39][40]. In
ED
addition the D and G bands, characteristic for disordered carbon systems, were present (this spectral region is better seen as a zoom in figure 8b)[Rob02]. The G-band at ~ 1600 cm-1 corresponds to the
PT
in-plane bond stretching movement of pairs of sp2 C atoms (all sp2 sites, not only in aromatic rings), while the D-band at ~ 1350 cm-1 is a breathing mode connected to the presence of sixfold aromatic
CC E
rings being activated by disorder [32][34]. The spectrum of the sample annealed at 200°C still showed a high photoluminescence background,
A
which was however lower than the value measured before annealing. This could indicate that some of the O-bridges and of the -OH groups were broken at 200°C and smaller aromatic parts remained, where the exciton was still localised on the phenol rings. In the case of amorphous hydrogenated carbon films, it was shown that the photoluminescence comes from isolated sp2 carbon clusters (as rings) and its intensity is lower when the hydrogen content decreases[41]. Our system annealed at these temperatures was different from the a-C:H films but still a system containing disordered 18
carbon. As one can see in the inset of Fig. 8a the intensity of the photoluminescence at 4200 cm-1 decreased strongly with increasing annealing temperature. This indicates that the sp2 carbon ring clusters were growing (becoming less isolated) and the system started to lose hydrogen with increasing temperature. Most likely in the 300-400°C temperature range the neighbouring C-OH groups were condensed. This induced on the one hand a decrease of the O and H content in the
IP T
system, and on the other hand clustering of different sp2 carbon aromatic parts. In a similar temperature range, using infrared spectroscopy it was shown that neighbouring Si-OH groups are
SC R
forming Si-O-Si bonds and water[42]. Formation of larger sp2 carbon aromatic clusters by breaking of
C-O bonds and -OH condensation was the mechanism proposed for synthesizing tannin-based Si codoped carbon for supercapacitor applications by a microwave-assisted method[43]. For our sample
U
the spectra in the Fig 8b (right panel) show the region of the -OH vibrations at ~3400 cm-1 with a clear
N
decrease of intensity for the sample which was annealed at 500°C. The spectra also show a broad
A
Raman band at ~2900 cm-1 which is a superposition of the C-H stretching vibrations with the second
M
order of the C-C vibration and thus difficult to interpret.
ED
Now we focus on the Raman spectra in the region of C-C vibrations. In nano-graphite systems, the decrease of the ID/IG intensity ratio is attributed to improved ordering and growth of the
PT
nano-crystals dimension. In contrast, in the case of more disordered amorphous carbon systems with small clusters, the interpretation of the ID/IG ratio is exactly opposite[32]. Thus, the increase of the D-
CC E
band and consequently of the ID/IG ratio indicates an increase of the size of the sp2 carbon clusters indicating ordering in the amorphous carbon system. The Figure 8b (left panel) shows the Raman spectra as a zoom of the region of the D- and G-bands. It should be mentioned that the D and G band
A
region in the Raman spectra of tannin-furanic rigid foams were interpreted by Reyer et al. in terms of rearrangement of furanic and tannic structures during polymerisation[38]. As can be seen, increasing the temperature for the Al2O3/Tannin/Lactose/Hexa/SiO2 samples from 200°C to 500°C induced an increase of the ID/IG ratio. For the ID/IG ratio (peak area ratio) values of 0.16, 1,17, 1,34 and 1,91 were obtained, by fitting the D-band at ~1387 cm-1 and G-band at ~1593 cm-1 using Gauss functions, for 19
the temperatures 200°C, 300°C, 400°C and 500°C, respectively. This increase of the ID/IG ratio with increasing annealing temperature may be associated to a growth of the sp2 carbon cluster area. This happened most likely because part of the OH groups from phenols were released due to thermal energy and the smaller aromatic molecular fragments connected forming larger aromatic sp2-carbon clusters. This behaviour is in very good agreement with the temperature evolution of the
increase of size of the sp2 carbon clusters with increase of the temperature too.
SC R
b)
ED
M
A
N
U
a)
IP T
photoluminescence which was previously discussed and indicated less hydrogen in the system and an
PT
Fig. 8: Raman spectra measured at room temperature of the Al2O3/Tannin/Lactose/Hexa/SiO2 samples annealed at 200°C, 300°C, 400°C and 500°C. The inset shows the value of photoluminescence at 4200 cm-1 as
CC E
a function of annealing temperature a) The same spectra zoomed in the regions of the D- and G- C-C vibrational bands and in the region of -OH stretching band. b)
A
Figure 9 shows the spectra measured at room temperature for annealed samples of
Al2O3/Tannin/Lactose/Hexa/SiO2 to which n-doped Si was added. In all spectra the Si phonon mode at 520 cm-1 and the D and G bands corresponding to C-C vibrations can be observed. If we compare the spectra to those in Figure 8, one can see that the photoluminescence intensity was lower for the samples mixed with n-doped Si. We speculate that n-Si can act as a catalyser of the clustering processes mentioned before. On the other hand, the lower photoluminescence intensity may also be 20
related to lower carbon content in the measured volume (part of the excited volume is now occupied by Si). However also in these spectra we observed an increase of the ID/IG ratio with annealing temperature, that was found by Ferrari et al. [32] to be "proportional to the number and clustering of rings", and indicates an increase of ordering in the amorphous carbon system. In the case of diamond-like carbon films where Si was co-deposited, increasing Si content was found to decrease
IP T
the ID/IG ratio, which was connected with a decrease of the sp2 bonded clusters and promotion of sp3 formation [44]–[47]. In our case the situation was different because n-doped Si was added to the surface
PT
ED
M
A
N
U
SC R
of the samples, so that an atomic mixing was not expected.
CC E
Fig. 9: Raman spectra measured at room temperature of the Al2O3/Tannin/Lactose/Hexa/SiO2 samples mixed with n-doped Si annealed at 200°C, 300°C, 400°C and 500°C. In order to better visualize the influence of n-Si addition to our samples, we directly
A
compared spectra measured at room temperature for the samples annealed with and without n-Si. Figure 10a shows the spectra measured for samples which have been annealed at 500°C and the figure 10b for a sample coked at 800°C. In both cases one can see the second order peak of Si at ~ 960 cm-1. The spectra are shown scaled to the same intensity of the D and G peaks. The spectra of the samples mixed with n-Si appeared to be noisier due to the smaller Raman excited carbon content, because part of the excited volume was occupied by Si. It is very clear for both annealing 21
temperatures that the addition of n-Si caused a decrease of the tail at ~1160 cm-1 and an increase of the signal at ~ 1470 cm-1 (indicated by arrows). Besides the known D- and the G-bands, the region of the C-vibrations generally shows additional features (at 1100-1200 cm-1 and 1400-1500 cm-1) which are lower in intensity and therefore are either neglected or have controversial interpretation in literature. Exactly in the region
IP T
were the n-Si caused changes in our spectra Ferrari et al. [48] and Piazza et al. [49] assigned Raman
bands at ~1150 cm-1 and at ~1460 cm-1 to transpolyacetilene segments. The band at lower energy
SC R
corresponds to a C-C vibration, while the band at higher energy to a C=C vibration in sp2 C chains in
transpolyacetilene [50]. In our complex carbon system, besides C rings also C chains may be present. The addition of n-Si to one sample clearly modified the carbon bonding. These two bands were found
U
to exhibit a strong dispersion with changing excitation energy [48]. Measuring Raman spectra with
N
different laser energy in order to check the dispersion of these modes could bring more information
M
A
regarding the C bonding in our system, but is beyond the aim of this paper. b)
A
CC E
PT
ED
a)
Fig. 10: Comparison of Raman spectra measured at room temperature for samples with and without n-Si annealed at 500°C (a) and coked at 800°C (b).
22
3.8 Impingement tests Fig. 11 shows the filters in the sand molds after the impingement tests. For every filter batch, three tests were carried out. All samples of AC5-2AH were destroyed when the molten steel was poured through the filters, #1. This result was expected, due to the very low CCS value of 0.05 MPa. The other three batches (AC5-2AHSi; AC5-2AH-alox; AC5-2AHSi-alox) survived the thermal shock and the
IP T
impingement itself. It was shown above that the addition of n-Si increased the CCS value. All AC5-
2AHSi samples were also intact after the experiment, #2 in Fig. 11. It is obvious that the Al2O3 coating
SC R
had a beneficial effect due to the higher CCS values. Therefore, the impingement test was positive
A
CC E
PT
ED
M
A
N
U
and the filters were also intact, #3 and #4 in Fig. 11.
Fig. 11: Foams in sand mold after impingement tests: 1=AC5-2AH; 2=AC5-2AHSi; 3=AC5-2AH-alox; 4=AC5-2AHSi-alox
23
4. Conclusions In the present work a new development of carbon-bonded alumina filters for steel melt filtration was presented. Samples were manufactured by replacing the conventional pitch binder Carbores®P, with a simple and environmental friendly system based on lactose and tannin. These two, unlike conventional binders derived from oil refining, are absolutely harmless and renewable. The lactose-
IP T
tannin binder system forms a resit network structure comparable to novolac resins. The lower
carbon yield compared to pitches resulted in filters with thinner struts and more cracks after firing.
SC R
This defects led to a relatively low cold crushing strength in comparison to the reference filters. The addition of n-Si increased the strength of the filters by improving the carbon yield of the binder system. An additional Al2O3 coating which sintered during heat treatment increased the CCS values
U
significantly. Raman spectroscopic measurements carried out at room temperature for
N
Al2O3/Tannin/Lactose/Hexa/SiO2 samples annealed at 200°C, 300°C, 400°C and 500°C showed a
A
reduced photoluminescence background and the ID/IG ratio increased with annealing temperature.
M
This behaviour may be explained by -OH thermal release and the formation of larger aromatic carbon
ED
ring clusters. The new filters were successfully tested by impingement with molten steel without
PT
preheating.
CC E
Acknowledgements
The authors would like to thank Ms. E. Qoku for X-ray diffraction measurements and Dr. G. Schmidt
A
for the SEM investigations. We also thank the German Research Foundation (DFG) for supporting these investigations within the Collaborative Research Centre 920 “Multi-Functional Filters for Metal Melt Filtration – A Contribution towards Zero Defect Materials”, subprojects A01, A04 and S03.
24
REFERENCES D. Janke, K. Raiber, Grundlegende Untersuchungen Zur Optimierung Der Filtration von Stahlschmelzen, 1996.
[2]
C.G. Aneziris, D. Borzov, J. Ulbricht, J. Suren, Cfi-Bericht DKG 2003, 80, E31.
[3]
C.G. Aneziris, D. Borzov, J. Ulbricht, J. Suren, in Unitecr, Osaka 2003, 631.
[4]
C.G. Aneziris, D. Borzov, J. Ulbricht, Interceramic Refract. Man. 2003, 22.
[5]
D. Borzov, J. Ulbricht, W. Schulle, in Int. Feuerfest-Kolloquium Aachen, 2001, 50.
[6]
Verordnung (EG) Nr. 1907/2006 (REACH): Registration, Evaluation, Authorisation and Restriction of Chemicals, 2006.
[7]
C. Biermann, Entwicklung Eines Neuen Umweltfreundlichen Bindemittel-Systems Für Die Feuerfestindustrie Auf Pharmazeutischer Und Lebensmittlechemicher Basis, TU Bergakademie Freiberg, 2016.
[8]
E. Gueguen, C.G. Aneziris, C. Biermann, Feuerfeste Formkörper Und Massen Sowie Bindemittel Und Verfahren Zu Deren Herstellung, 2017, WO 2017118449 A1.
[9]
M. Hampel, C.G. Aneziris, Cfi-Bericht DKG 2007, 84, E125.
[10]
M. Hampel, C.G. Aneziris, Tech. Keramische Werkstoffe 2007.
[11]
A. Gardziella, J. Suren, Stahl Und Eisen 1999, 119, 135.
[12]
A. Gardziella, J. Suren, Stahl Und Eisen 1993, 113, 75.
[13]
C.G. Aneziris, J. Hubálková, R. Barabás, J. Eur. Ceram. Soc. 2007, 27, 73.
[14]
U. Klippel, Interceramic Refract. Man. 2006, 1, 14.
[15]
U. Klippel, in 49. Int. Feuerfest-Kolloquium, Aachen 2006, 6.
[16]
U. Klippel, V. Stein, C.G. Aneziris, in UNITECR ’07, 2007.
[17]
V. Stein, C.G. Aneziris, U. Klippel, W. Schönwelski, E. Guéguen, in Unitecr, Salvador 2009.
[18]
S. Zhang, N.J. Marriott, W.E. Lee, J. Eur. Ceram. Soc. 2001, 21, 1037.
[19]
S. Zhang, W.E. Lee, J. Eur. Ceram. Soc. 2001, 21, 2393.
[20]
W.E. Lee, S. Zhang, Int. Mater. Rev. 1999, 44, 77.
[21]
A. Yamakuchi, Taikabutsu Overseas 1999, 4, 14.
[22]
A. Watanabe, H. Takahashi, S. Taganaga, Taikabutsu Overseas 1987, 4, 17.
A
CC E
PT
ED
M
A
N
U
SC R
IP T
[1]
[23]
A. Yamaguchi, S. Zhang, J. Yu, J. Am. Ceram. Soc. 1996, 79, 2509.
[24]
V. Stein, Contribution to the Characteristic Improvement of Carbon Bonded Doloma Refractories by Addition of Functional Ceramic Materials, TU Bergakademie Freiberg, 2011.
[25]
V. Stein, C.G. Aneziris, J. Ceram. Sci. Technol. 2014, 5, 115.
[26]
J.E. Lang, A. Karl, H. Rauleder, E. Müh, G. Stochniol, Verfahren Zur Pyrolyse von Kohlenhydraten, 2010, WO 2010/037699 A2.
[27]
M. Emmel, C.G. Aneziris, Ceram. Int. 2012, 38, 5165. 25
J. Luyten, S. Mullens, J. Cooymans, A.M. De Wilde, I. Thijs, R. Kemps, J. Eur. Ceram. Soc. 2009, 29, 829.
[29]
K. Schwartzwalder, A. V Somers, Method of Making Porous Ceramic Articles, 1963, 3,090,094.
[30]
M. Emmel, C.G. Aneziris, J. Mater. Res. 2013, 28, 2234.
[31]
M.A. Tamor, W.C. Vassell, J. Appl. Phys. 1994, 76, 3823.
[32]
A.C. Ferrari, J. Robertson, Phys. Rev. B 2000, 61, 14 295.
[33]
C. Casiraghi, A.C. Ferrari, J. Robertson, Phys. Rev. B - Condens. Matter Mater. Phys. 2005, 72, 1.
[34]
G. Irmer, A. Dorner-Reisel, Adv. Eng. Mater. 2005, 7, 694.
[35]
M. Emmel, Development of Active and Reactive Carbon Bonded Filter Materials for Steel Melt Filtration, TU Freiberg, 2014.
[36]
A. Weidner, D. Krewerth, B. Witschel, M. Emmel, A. Schmidt, J. Gleinig, O. Volkova, C.G. Aneziris, H. Biermann, Steel Res. Int. 2016, 87, 1038.
[37]
B. Luchini, J. Grabenhorst, J. Fruhstorfer, V.C. Pandolfelli, C.G. Aneziris, J. Am. Ceram. Soc. 2018, 1.
[38]
A. Reyer, G. Tondi, R.J.F. Berger, A. Petutschnigg, M. Musso, Vib. Spectrosc. 2016, 84, 58.
[39]
T.H. Huang, C.C. Hsu, C.T. Kuo, P. Lu, W.S. Tse, D.P. Wang, T.C. Chou, A.Y.G. Fuh, J. Appl. Phys. 1994, 75, 3599.
[40]
C. Röder, T. Weißbach, C. Himcinschi, J. Kortus, S. Dudczig, C.G. Aneziris, J. Raman Spectrosc. 2014, 45, 128.
[41]
B. Marchon, Jing Gui, K. Grannen, G.C. Rauch, J.W. Ager, S.R.P. Silva, J. Robertson, IEEE Trans. Magn. 1997, 33, 3148.
[42]
C. Himcinschi, M. Friedrich, K. Hiller, T. Gessner, D.R.T. Zahn, Semicond. Sci. Technol. 2004, 19, 579.
[43]
S.K. Ramasahayam, U.B. Nasini, A.U. Shaikh, T. Viswanathan, J. Power Sources 2015, 275, 835.
[44]
J.F. Zhao, P. Lemoine, Z.H. Liu, J.P. Quinn, J.A. McLaughlin, J. Phys. Condens. Matter 2000, 12, 9201.
[45]
P. Papakonstantinou, J.. Zhao, P. Lemoine, E.T. McAdams, J.A. McLaughlin, Diam. Relat. Mater. 2002, 11, 1074.
[46]
G.A. Abbas, J.A. McLaughlin, E. Harkin-Jones, Diam. Relat. Mater. 2004, 13, 1342.
[47]
K. Er, M. So, J. Ceram. Process. Res. 2011, 12, 187.
A
CC E
PT
ED
M
A
N
U
SC R
IP T
[28]
[48]
A.C. Ferrari, J. Robertson, Phys. Rev. B 2001, 63, 121405.
[49]
F. Piazza, A. Golanski, S. Schulze, G. Relihan, Appl. Phys. Lett. 2003, 82, 358.
[50]
E. Mullazzi, G.P. Brivio, E. Faulques, S. Lefrant, Solid State Commun. 1983, 46, 851.
26