BOLETIN DE LA SOCIED AD ESPAÑOLA DE Cerámica y Vidrio

BOLETIN DE LA SOCIED AD ESPAÑOLA DE Cerámica y Vidrio

Chemical interaction silicon nitride ceramics and iron alloys
F.J. OLIVEIRA, R.F. SILVA, J.M. VIEIRADepartment of Ceramics and Glass Engineering, UIMC, University of Aveiro, 3810-193 Aveiro, Portugal
Metal/ceramic diffusion experiments are helpful to study bonding mechanisms or the effect of metal composition on the chemical wear of ceramic cutting tools. The reaction kinetics of Fe alloys/Si3N4 ceramic diffusion couples was investigated in the temperature range 1050ºC-1250ºC, for 0.5h to 80h, under inert atmosphere. Optical microscopy, SEM and EPMA were carried out in cross sections of the reacted pairs. Si3N4 decomposes into Si and N that dissolve and diffuse through the metal. Both the diffusion zone on the metal side and the reaction zone on the ceramic side obey parabolic growth laws of time, with activation energies in the range Q=310-460kJmol-1. The amount of dissolved Si, the length of the diffusion zone and thus the reactivity of the ceramic increase as the alloy carbon content decreases. Due to Si accumulation, the α-Fe solid solution is stabilised at the reaction temperature and a steep decrease in the Si concentration is observed beyond the diffusion zone. The reinforcement of the Si3N4 composites with A12O3 platelets enhances the chemical resistance of the ceramic due to the inert- ness of this oxide and to the partial crystallisation of the intergranular phase. Other dispersoids such as HfN, BN and TiN do not improve the chemical resistance of the matrix by iron attack.Keywords: silicon nitride, steels, diffusion couples, interface kinetics
Interacción químico entre nitruros de silicio y aleaciones de acero
Los experimentos de difusión metal/cerámica permiten estudiar mecanismos de unión y analizar el efecto de la composición del metal en el desgaste químico de herramientas de corte cerámicas. En este trabajo se investigó la cinética de reacción en pares de difusión aleaciones de Fe/Si3N4 a temperaturas entre 1050ºC-1250ºC, tiempos entre 0.5h a 80h, en atmósfera inerte. Las secciones transversales de los pares de difusión se analizaron mediante microscopía óptica, SEM y microsonda electró- nica. El Si3N4 se descompone en Si y N que se disuelven y difunden en el metal. Tanto la zona de difusión en el metal como la zona de reacción en la cerámica obedecen una ley parabólica de crecimiento, con energías de activación de 310-460KJ.mol-1. La cantidad de Si disuelto, el tamaño de la zona de difusión y, por lo tanto, la reactividad de la cerámica aumenta al diminuir el contenido de carbón de la aleación. Debido a la acumulación de Si, la solución sólida de a-Fe se estabiliza a la temperatura de reacción, y se observa un descenso significativo en la concentración de Si más allá de la zona de difusión. El reforzamiento del Si3N4 con plaquetas de Al2O3 aumenta la resistencia química del Si3N4 debido a la inercia de este oxido y a la cristalización parcial de la fase intergranular. La incorporación de HfN, BN y TiN no mejoran la resistencia química de la matriz al ataque por Fe.Palabras clave: nitruro de silicio, aceros, pares de difusión, cinética de interfases
1. INTRODUCTION
Silicon nitride (Si3N4) ceramics are currently being used as wear parts and engine components. For a successful use in applications that may have intricate geometry, bonding to a different material or to itself is often a requirement [1,2].Brazing can be done at temperatures below 900ºC, thus limi- ting the maximum useful temperature of the component [3]. Direct bonding to refractory metals or alloys of higher melting points, such as Fe or Ni base alloys are alternative methods still under study [4]. The interactions between iron alloys and Si3N4 ceramics are of importance also in the field of cuttingtools [5-7]. It is well known that silicon nitride cutting toolswear rapidly in high speed machining of most steels, whilebeing able of machining grey cast iron with minimum wearrates [8]. Kramer and Suh [9] have proposed a model wherethe solubility of the tool material in pure iron determines thewear rate relative to HfC. Vleugels et al. [10] developedKramer´s model to account for the effect of the iron alloy com-position on the wear rate of silicon nitride cutting tools. In thismodel, the chemical affinity between alloying elements andthe nitrogen determines the reactivity of a given alloy. Silva
Bol. Soc. Esp. Cerám. Vidrio, 39 [6] 711-715 (2000)
and others [7] further improved the theoretical approach, considering the chemical cross effects of alloying elements on both Si and N originating from ceramic decomposition.These approaches are established on a thermodynamic basis and further improvements are needed concerning the reaction kinetics. Solid state diffusion couples have been used as way of determining the reaction rates of Si3N4 or SiAlONceramics in contact with Fe alloys [2,10]. The study of interac-tions between the Si3N4 ceramics and the iron alloys should bring about the effects of alloys compositions and ceramic sin- tering additives on the reaction mechanisms. The incorporation of nitride and oxide compounds of large negative Gibbs energyof formation also affects the overall chemical resistance of the Si3N4 ceramics in the contact with iron alloys [11]. In the pre- sent work, the reactivity of several Si3N4 monolithic and com- posite ceramic materials in diffusion couples with pure Fe, carbon and chromium alloyed steels is investigated. Equilibrium thermodynamic calculations at moving interphase boundaries in the metal are tentatively correlated to concentration profiles of Si diffusing from the ceramic side of the couple. 2. EXPERIMENTAL DETAILS
Metal/ceramic diffusion couples, produced under inert atmosphere or vacuum, were used to study the reaction kine- tics at temperatures in the range 1050ºC-1250ºC for dwelling times of 0.5h to 80h. The experimental set up and the geome- try of the diffusion couples have been described in previous publications [12,13,14]. A thin piece of ceramic (60-150µm) is placed between two larger steel slabs that forge around the ceramic at the reaction temperature. Alumina platelets (Grade T2, ELF ATOCHEM) were placed between the ceramic and the steel to mark the initial contact plane. The couples are supported with alumina spacers in a graphite jig, inside an alumina tube with water cooled caps. The load is applied using a cantilever to values between 5MPa and 7.5MPa.Several silicon nitride (Si3N4) based ceramics were develo- ped (Table I) to study the effect of matrix composition and of reinforcing phases on the reactivity in contact with iron alloys

The Si3N4 monolithic material (SN1) was fully densified by hot-pressing at 1650ºC/30MPa/60min. A somewhat hig- her temperature (1700ºC/120min/30MPa) was necessary for densification of the composites with the SN1 matrix to over-come the constraining effects of the inclusions upon sinte- ring. The composite containing the alumina platelets (SN2- AL) was developed aiming a reduction of dissolution of the reinforcing phase, by saturating with A12O3 the intergranu-lar glassy phase always present in silicon nitrideceramics [16]. This composite was hot-pressed at1500ºC/90min/30MPa. All the ceramics were hot-pressed ina BN coated graphite die, thermally insulated with coarsealumina powder, and heated in air by a radio frequencyinduction generator.The study of the effect of iron alloys composition on thereaction kinetics was done with pure iron and five commer-cially available steels with different amounts of carbon andchromium (Table II). The iron alloys are three carbon steelswith increasing amount of carbon, A1, A2 and A3 and twochromium containing steels, AC1 and AC2 where AC1 hasless carbon than AC2.After reaction, the metal/ceramic interface was cross-sec-tioned and polished for observation and chemical analysis byscanning electron microscopy (SEM- Hitachi S4100) and elec-tron probe microanalysis (EPMA -Cameca). Calibration of theEPMA with Cr, Fe and Si standards was made prior to theanalyses while the SEM/EDS system used software calibra-tion methods. Although detectable, carbon and nitrogen werenot quantified in either of these two systems, due to contami-nation in the case of carbon, and due to the low amount ofnitrogen that was below the detection limit of the EPMA(about 0.04wt%). The length of the affected zones on both theceramic and the iron alloys was evaluated using an opticalmicroscope (Zeiss) and the image analysis software Quantimet500+ (Leica Cambridge). a Fe: Aldrich, 99,98% purity ( 65ppm Na, 60ppm Si, 25ppm Al, 5ppm Ti, 3ppmCa, 1ppm Mg)

3. RESULTS AND DISCUSSION
Silicon nitride decomposes in contact with the iron alloys for all the experimental conditions tested in this work, the Si and N diffusing into the metal, leading to a diffusion zone of variable thickness, d, on the metal side of the couples (Figure1a). EPMA analyses evidenced that Si remains in the metal adjacent to the ceramic while no N was detected. The low nitrogen partial pressures in the reaction chamber (0,5Pa for the Ar N50, Ar Líquido) and the low solubility of nitrogen in Fe at these temperatures [17] give rise to an easy escape of N to the atmosphere. The effect of nitrogen partial pressure on the reactivity was already assessed by others [18,19], the reac- tivity being highest for the lowest nitrogen partial pressure.The additives of SN1 remain as oxides or mixed oxides in a zone of thickness r corresponding to the consumed ceramic (Figure 1a). The position of the initial contact plane, K in Figure 1a, between the ceramic and the alloy is marked with alumina platelets. The boundaries between this modified metal/ceramic contact zone and the unaffected ceramic and steel regions are also clearly visible in Figure 1a.The microstructure of the diffusion zone d depends on the composition of the iron alloy. Figures 1b and 1c correspond to diffusion experiments between SN1 ceramic and two steels of different carbon contents, A1 (0.13wt% C) and A3 (1.05wt% C). After etching with an ethanol/1.5% HNO3 solution (Nital1,5%), the microstructure of the diffusion zone of the hypo-eutectoid steel A1 shows that there is no pearlite present. Thismeans that the amount of dissolved Si is enough to stabilisethe α−Fe solid solution at the reaction temperature over thewhole length of the diffusion zone. Pearlitic grains appear

Figure 1 – SEM photomicrographs of diffusion couples of SN1 cera- mic after reaction at 1050ºC/80h against (a,b) A1 carbon steel; (c) A3 carbon steel, (d) AC2 chromium alloyed steel. d – diffusion zone in the metal side, r – reaction zone in the ceramic side.

Figure 2 – Effect of steel composition on the diffusion zone forma- tion rate for SN1 ceramic couples with pure iron (Fe) and different steels (A1, A2, A3, AC1). (a) Square of the diffusion zone thickness d at 1150ºC as a function of time; (b) Arrhenius plot of the parabolic diffusion rate constant Kd for the temperature range 1050-1250ºC. only in the unaffected metal after the diffusion zone (Figure1a) resulting from the decomposition of the γ-Fe solid solutionduring cool down to room temperature. On the contrary, theA3 steel keeps its pearlitic structure in the diffusion zone(Figure 1c), because the much higher carbon amount hindersthe γ-Fe→α-Fe transformation.For steel AC2, the very high carbon content also stabilisesthe γ-Fe phase at the reaction temperature in spite of the pre-sence of the Cr from the steel and Si from the Si3N4 ceramic decomposition. Both are α-Fe stabiliser elements. In this par- ticular case, large (Cr,Fe)7C3 carbide particles are seen in a pearlitic matrix (Figure 1d).The thickness of the diffusion zone, d, or extent of the α-Fe solid solution domain, was measured as a function of time and is reported in Figure 2a for pure iron and the three carbon steels A1, A2 and A3. A parabolic law of time for d, d2=2Kd.t [20], fits the experimental data. Since N diffuses much fasterthan Si in both iron structures [21,22] the extent of the diffu- sion zone must be related to the Si concentration profile. The amount of Si dissolved within d zone is a measure of the reac- tivity of the Si3N4 ceramic due to the direct relationship bet- ween d and the thickness of the ceramic reacted zone, r [23],significantly decreases with the increasing nominal carbon content of the steel. This is also true for the chromium contai- ning steels that have similar amounts of Cr but quite different C content [23].The activation energy for the parabolic rate constant Kd increases from 310kJmol-1 for pure iron to 360kJmol-1 for A1 carbon steel and to about 460kJmol-1 for the Cr alloyed steel AC1. These values calculated from the Arrhenius plots in Figure 2b, are comparable to the activation energies measuredfrom similar plots for the reaction zone (320→440kJmol-1) [12]. Stoop and co-author [2] calculated an activation energy of700kJmol-1 in AISI 316 stainless steel/Si3N4 couples for the diffusion zone kinetics. This high value was attributed to a mixed control for the reaction, the ceramic decomposition and the diffusion in the metal. The values found in the presentwork are higher than the activation energy for Si diffusion in α-Fe (240kJmol-1) [22], but closer to 400kJmol-1, a value repor- ted for the Si3N4 densification kinetics [24] and for the Si-N bond energy [25]. This is a first indication that the ceramicdecomposition, dissolution and diffusion in the intergranular glassy phase are among the slowest steps of the overall reac- tion kinetics.The Si concentration profiles measured with the EPMA in the diffusion zone of couples with SN1 against pure iron, A1 and A3 steels, after reaction at 1150ºC for 7.5 hours, are repre- sented in Figure 3a. The integral area under the curves, the amount of dissolved Si, increases in the sequence A3→A1→Fe corresponding to the enhanced reactivity for the alloys with lower carbon content already observed in Figure 2a. The pro- file shape of Si on the A3 steel differs from those of the low carbon alloy A1 and pure Fe, these two showing a steep decrease on the Si concentration in a plane corresponding to the interface between d and the unaffected metal. The last feature was already reported by Heikinheimo [18] and Oliveira et al. [26] in diffusion couples with pure iron and it was attributed to the presence of the α-Fe/γ-Fe phase boun- dary.Taking this as a valid assumption we calculated the com- position at the α-Fe/γ-Fe solid solution phase boundary for the various steels [23]. The equilibrium thermodynamic calcu- lations at the advancing interfaces on the metallic side of the diffusion couples were done using the ChemSage program [27] (V. 3.2, GTT Technologies, RWTH) and the SGTE [28] databases for pure substances and solutions. The calculation method used, the assumptions made and the results of ther- modynamic modelling are described in another publication [23] where a direct correlation was obtained between the reac- tivity and the concentration of Si at the α/γ-Fe phase boun- dary. The calculated Si concentrations necessary to fully stabi- lise the α-Fe solid solutions are given in Figure 3a for Fe, A1 and A3 steels. These values determine the position of the abrupt decrease in the Si concentration profile and thus the extent of the diffusion zone. This is verified for the low carbon steel A1 and pure iron whereas for the high carbon content steel A3, the Si incorporated was not enough to promote the γ-Fe to α-Fe phase transformation (Figure 3a). In this alloy, Si diffusion occurs only in the close packed fcc structure (γ-Fe) where diffusion coefficients of substitutional elements are the lowest [22].The Si concentration profiles in the diffusion zone, d, of SN1/pure Fe couples reacted at increasing times, are shown in Figure 3b. The value of Si concentration at which the sudden decrease in the profile occurs is the same, irrespecti- vely of the degree of reaction, further confirming the validity of the calculations made.The effect of the silicon nitride composition on the reac- tion kinetics was studied in diffusion couples with the steel A1. The compositions of the tested ceramic matrix composites are given in Table I. The set of SEM micrographs presented in Figure 4 compares the microstructures of the diffusion couples after reaction at 1150ºC for 20h. The HfN composite (Figure4a) is more reactive than the composite with BN (Figure 4b), or TiN (Figure 4c). The more chemically resistant material is the composite SN2-AL with A12O3 platelets dispersed in an alumina saturated matrix (Figure 4d). The morphology of thereaction zone results from the decomposition of Si3N4 and filling up with Fe the space between the reinforcing particles. The lengths of the diffusion and reaction zones are also inte- rrelated as discussed before for the couples with the un-rein-forced ceramic SN1. The larger diffusion zone is observed for the couple with the more reactive ceramic, SN1-HN (Figure4a) while the smallest value of d is measured in the couples with the ceramic composite SN2-AL (Figure 4d).The extent of the reaction zone, r, for these composites in the reaction with carbon steel A1 at 1150ºC is plotted as a function of time in Figure 5. Data for SN1 are also included for compari- son. As it happened with the diffusion zone of Figure 2a, a parabolic law fits the experimental data. The compounds TiN, BN, HfN and A12O3 should be chemically more stable thanSi3N4 in the contact with iron at high temperature due to themore negative free energy of formation of these nitrides and ofthe Al oxide [9]. However, as depicted from Figure 5, only theSN2-AL composite with the A12O3 platelets reveals increased resistance to chemical attack by the carbon steel A1, as confir- med by the smaller penetration depth, r, into the composite than that into the SN1 matrix. The composites with the compound BN and particularly the one with HfN evidence faster reactionrates than the values obtained for the SN1 matrix or the SN1-TN composite. While TiN particles were inert with respect to the unreinforced matrix during the sintering stage, the BN and HfN reacted to some extent with the intergranular glassy phase [14], increasing the area fraction for Fe diffusion into the ceramic. In the case of the SN2-AL composite, the matrix contains X-SiAlON from the reaction between the added A12O3 and Si3N4 [16], thestability of the crystalline intergranular phase delaying thecorrosion of the ceramic by Fe [29].

Figure 3 – Si concentration profiles in the diffusion zone of SN1/ steel couples tested at 1150ºC. (a) Si concentration in iron, A1 and A3 carbon steels, after reaction for7.5 h; (b) values for pure iron for different reaction times.
Accordingly to these findings we propose that the attack proceeds by Fe penetration through the grain boundary phase into the ceramic followed by the dissolution of the Si3N4 grains in the modified glassy phase. The smaller the amount of theintergranular glassy phase the smaller the reaction rates are. The activation energy for the formation of the reaction zone in this composite was calculated as 460kJmol-1 [12], a value com- parable to the given above for the couples with SN1.4.

CONCLUSIONSSi3N4 decomposes into Si and N that dissolve and diffuse in pure iron, carbon and chromium alloyed steels at tempera- tures above 1050ºC. Si remains in the alloy diffusion zone adjacent to the ceramic while N escapes to the atmosphere.For low carbon alloys, the dissolved Si is enough to stabilise the α-Fe solid solution at the reaction temperature. In these alloys a sudden decrease in the Si concentration is observed in the plane corresponding to the α-Fe/γ-Fe phase transition. Steels with low carbon content also increase the amount of Si dissolved, the length of the diffusion zone and thus the reac- tivity of the ceramic.

Figure 4 – SEM photomicrographs of diffusion couples of composite ceramics after reaction at 1150ºC/20h with A1 carbon steel. (a) Si3N4–HfN composite; (b) Si3N4–BN; (c) Si3N4–TiN; (d) Si3N4–Al2O3. d – diffusion zone in the metal side, r – reaction zone in the ceramic side. Figure 5 – Effect of ceramic composition on the reaction zone (r) for- mation rate for diffusion tests with A1 carbon steel at 1150ºC.
The Si3N4 composites containing 30vol% of HfN or BN did not show any improvement in the chemical resistance relatively to the unreinforced material, due to unwanted reac- tivity during the sintering stage between those compoundsand the intergranular phase of the matrix. For the TiN-Si3N4 composite the reaction rate is the same as for the matrix while for the composite with the A12O3 platelets the chemical resis- tance is enhanced due to the presence of stable crystalline phases as intergranular phases.Both the diffusion and reaction zones obey a parabolic growth law of time with activation energies for the correspon- ding rate constants in the range Q=310-460kJmol-1. The acti- vation energies values are closer to values known for the control by diffusion mechanisms of hot-pressing, oxidation or corrosion kinetics in the ceramic than to the values of Si diffu- sion in the iron alloys.
ACkNOwLEDGMENTSF. J. Oliveira acknowledges the financial support of theSub-Programa Ciência e Tecnologia do 2º Quadro Comunitáriode Apoio. The financial support under the research contractPRAXIS 3/3.1/MMA/1777/95 is gratefully acknowledged.

IBAÑEZ JESUS
CRF
FUENTE: http://boletines.secv.es/upload/20090514113342.200039711.pdf