Kozo Ishizaki and Koji Matsumaru
Presented at the 2011 International Conference on Hot Isostatic Pressing Kobe, Japan, 12-14 April 2011
Submitted: 12 April 2011, Accepted: 24 May 2011
Topics CoveredAbstractKeywordsIntroduction Silicon Nitride Aluminium Nitride Borosilicate GlassConclusionsReferencesContact Details
Ellingham diagrams are very powerful phase diagrams to evaluate equilibrium phases under gas-solid reactions. Those phase diagrams were extended to higher pressures to examine stable phases under high total gas pressures. The variables under consideration consist not only of temperature and partial pressure of a gas involving the chemical reactions, but also total gas pressures. These new phase diagrams are applied to predict stable phases under hot isostatic pressing conditions (HIP phase diagram). Silicon nitride sintering, aluminium nitride sintering, oxide super conductors sintering and borosilicate glass melting were discussed as examples of HIP phase diagram.
Phase Diagram, Ceramics, Sintering, Ellingham Diagram, Gas Pressure, Gas-Solid Reaction
Thermodynamics can provide important criteria for technical problems. Materials processing involves many complicated reactions. Therefore, thermodynamics in materials engineering is important, especially ceramics processing under high gas pressures. There are many unanswered questions in HIP processes for ceramics. For example, why HIPed ceramics have different colors sometimes. Normally Ellingham diagrams are considered independent to the total gas pressure. The authors, however, describe a generalized method to use Ellingham diagrams under high gas pressures (HIP phase diagram) for answering these questions of gas-solid chemical reactions under high gas pressures.
Sintering problems of silicon nitride are mainly oxide phase and carbon impurity. The behavior of carbon was investigated as an impurity of HIP sintered silicon nitride. Carbon coated silicon nitride particles were mixed with carbon free particles at a given ratio to control the carbon proportion, and the mixed particles were sintered by the glass capsuled HIP process as well as hot pressed. The results for carbon contents after sintering are shown in Fig.1 for silicon nitride powder carbon contaminated up to 4000 ppm . Carbon contents changed little by HIP sintering, on the other hand, carbon decreased to almost a constant value for hot pressed samples. SiC was found in hot pressed samples, but graphite in HIPed samples .
Figs. 2 (a), (b) and (c) show parts of the HIP phase diagram for the oxidation of silicon, carbon and silicon carbide at (a) 0.1, (b) 60 and (c) 200 MPa. The partial pressure of oxygen in the presence of carbon moves along the 2C + O2 = 2CO line. This line intersects with the SiC + O2 = SiO2 + C line at point P. In the region where the oxygen partial pressure is higher than those of point P and Q, SiC,CO and SiO2 are stable. The temperature at point P increase from 1600 to 2150 and 2350 K as the total gas pressure increased. In hot pressing processes, the gas pressure is around 0.1 MPa, and hence the region where carbon monoxide is stable (the region above the 2C + O2 = 2CO line) is wide. Carbon is oxidized to form carbon monoxide and the quantity of total carbon is reduced, and silicon carbide is formed during the sintering. On the other hand, the HIP sintering process has a wider stable carbon region. Therefore, the carbon added to the starting silicon nitride powder is stable as solid phase carbon, i.e., graphite.
One has to pay particular attention that the reaction, 2C + O2 = 2CO has one degree of freedom more than normal gas (oxygen)-solid reactions. Because oxide CO is a gas phase and not form another phase. This is the reason why carbon can react unexpectedly under almost any conditions.
Figure 1. Carbon amount of sintered silicon nitride. HIP A: HIP treated sample at 1823 K and 60 MPa for 1 h, HIP B: HIP treated sample at 1973 K and 60 MPa for 1 h, HP: hot pressed sample at 1873 K and 20 MPa for 1 h.
Figure 2. Parts of HIP phase diagram at (a) 0.1, (b) 60 and (c) 200 MPa
Aluminium nitride has excellent thermal and mechanical propertiers. The AlN thermal conductivity decrease by impurities such as oxygen, silicon, iron and magnesium . Especially, oxygen atoms form an oxynitride spinel phase and / or an AlN pseudo-polytype phase which reduces greatly the thermal conductivity of sintered AlN . Therefore, it is important to understand the oxygen behavior to evaluate the thermal conductivity.
AlN powder with 1 mol% Y2O3 powder were sintered by HIP and normal furnace. Fig.3 shows the oxygen contents of sintered AlN at various sintering temperatures . The oxygen contents of HIPed samples are almost constant. On the other hand, the oxygen contents of the normally sintered samples linearly decrease as the sintering temperature increase.
The expected oxygen contents, which influence the AlN thermal conductivity, are evaluated by using the HIP phase diagram for the oxidation reactions of carbon, the related material and AlN. Considering the reaction 2C + O2 = 2CO in a graphite furnace for the normal sintering, the oxygen partial pressure is assumed to be along a line corresponding to this reaction. It is assumed that oxygen atoms exist as Al2O3 on the particle surface of AlN powders.
For normal sintering with carbon heater the following reaction
2/3Al2O3 +2/3N2 + 2C = 4/3AlN + 2CO (1)
may proceed. Another reaction may be considered.
2/3Y2O3 +2/3N2 + 2C = 4/3YN + 2CO (2)
Fig.4 shows a part of the HIP phase diagram . Reactions at normal pressure are plotted by dashed lines and those at 60 MPa of the total pressure are plotted by solid lines. Considering of equations (1) and (2) for the normally sintered AlN, the oxygen contents of the samples could be decreased. Yttrium nitride should be produced by equation (2) after sintering at around the point R, but was not found in the sintered bodies by examining X-ray diffraction. Fig.4 indicates that the points O, P, Q and R shifts to O60, P60, Q60 and R60 in accordance with the increase of the gas pressure. HIP sintering temperature was lower than the points P60 and Q60. Therefore, reactions of equations (1) and (2) will not proceed in the HIP process and oxygen will react to form an oxide under 60 MPa gas pressure.
Figure 3. Oxygen contents of the sintered AlN as a function of sintering temperature.
Figure 4. A part of the HIP phase diagram for the related reactions of sintering AlN.
Superconducting ceramics (BiSrCaCu2Ox and BiPbySrCaCu2Ox) were obtained directly by capsule-HIP densification without any other additional treatment for the first time by the author’s group . An interesting aspect of this material is the so-called “partial melting.” The high-Tc phase is usually obtained just below the “partial melting” point, at which temperature some of the constituent substance melt [7-8]. There is strong influence of Po2 on this partial melting. This does not imply a pressure effect on melting point after the Clausius-Clapeyron equation, but that some of the substances may be reduced and then melt. Assuming this partial melting phenomenon is a reduction reaction, a part of HIP phase diagram is drawn by using the experimental data of capsule-free O2 HIP for 100 MPa and reported data [7-8] for 0.1 MPa as in Fig. 5. Their partial melting line under 0.1 MPa draws as the dashed line, where squares, triangle and circles indicate melted, partially melted and not melted samples, respectively. The data of Endo  and Endo  are indicated by open and half-open marks, respectively. The partial melting line under 100 MPa of a total gas pressure draws as the dashed and dotted line. The experimental data for capsule-free O2 HIPing are plotted as solid marks. The high total gas pressure affects the formation line of partial melting.
Figure 5. The HIP phase diagram for the formation of high-Tc phase of Bi-Sr-Ca-Cu-O superconductor material with partial pressure lines of oxygen.
The high-temperature oxidation reaction of carbon is very important because of common carbon heaters in HIP equipment or graphite crucible used for powder container. Possible chemical changes of the carbon have a dramatic effect on the overall microstructure and mechanical porosity of the borosilicate glasses materials. The structure of interfacial region between HIPed borosilicate glasses and carbon oxides which is affected by the gas pressure can be very complex, not only depending on specific reaction conditions such as, temperature, partial pressures of the gaseous species and matrix composition, but also total gas pressure.
Figs. 6 (a) and (b) show a photograph of HIPed borosilicate glass samples. It was shown that the borosilicate glass surface color was not uniform and changed from black to gray when gas pressure form 7 to 30 MPa. When gas pressure was reached 40 MPa, the surface color became uniform and white. Photographs of cross surface of borosilicate glasses HIPed at 1100°C are shown in Fig. 6 (b). The thickness of the carbon-rich dark layer decreased as the total gas pressure increased form 7 to 40 MPa.
In order to investigate the formation reactions for carbon-rich layer on the surface of borosilicate glass materials, the values of standard free energy were calculated for the reactions of graphite crucible under normal pressure.
The possible carbon oxidation reactions are
C +O2 = CO2 (3)
2CO + O2 = 2CO2 (4)
2C + O2 = 2CO (5)
The HIP phase diagram of the related reactions at the pressure 40 MPa are plotted in Fig. 7 by the solid lines. Point O is intersection of the reactions (3) (4) and (5). The figure indicates that the equilibrium point O0.1 (708 °C) shifts to O40 (1104 °C), as well as O100 (1195 °C) and each oxidation reaction line also shifts toward higher temperatures in accordance with the increase of the total-gas pressure to 40, and 100 MPa respectively. The stable region of carbon in the figure widens, and the stable region of CO gas reduces with the increase of the gas pressure. HIP sintering temperature 1100 °C was nearly equal to the point O40. Therefore, in the case of the normal sintering with the total gas pressure lower than 40 MPa and temperature at 1100 °C, there was a larger stable region of CO gas in the phase diagram. The reactions of equations (4) and (5) can proceed in the HIP process, and the graphite crucible is oxidized to generate CO gas. The CO gas generated is released into the sintering furnace, and the reaction between CO gas with borosilicate glass, especially on the surface of glass will take place.
Figure 6. Photograph of HIPed borosilicate glass at 1100°C at different gas pressure. (a) Topview, (b) Cross section view.
Figure 7. A part of the HIP phase diagram of the related reactions.
HIP phase diagrams proposed in this work to clarify many HIP sintering problem which were not possible to be solve before. Using the proposed HIP phase diagrams, silicon nitride sintering, aluminium nitride sintering, oxide superconductor densification and borosilicate glass melting are explained and agreed well with the experimental results.
1. K. Watari and K. Ishizaki, “Influence of Gas Pressure on HIP Sintered Silicon Nitride and Stability of Carbon Impurity”, J. of Ceram. Soc. Jpn., 96 (1988) 535-540.
2. K. Watari, K. Ishizaki and M. Kawamoto, “Evaluation of Carbon Behavior in HIP’ed Silicon Nitride”, J. of Ceram. Soc. Jpn., 96 (1988) 741-748.
3. N. Kuramoto, H. Taniguchi, Y. Nomura and I. Aso, J. Ceram. Soc. Jpn, 93 (1985) 517.
4. T. Sakai, M. Kuriyama, T. Inuka and T. Kijima, J. Ceram. Soc. Jpn, 86. 30 (1978)
5. K. Ishizaki and K. Watari, “Oxygen Behavior of Normal and HIP Sintered AlN“, J. Phys. Chem. Solids., 50  (1989) 1009-1012.
6. H. Seino, K. Ishizaki and M. Takata, “HIPed High Density Bi-(Pb)-Sr-Ca-Cu-O Superconductors Produced without Any Additional Treatment”, Jpn. J. of Appl. Phys., 28 (1989) 78-81.
7. U. Endo, S. Koyama and T. Kawai, “Preparation of the High Tc phase of Bi-Sr-Ca-Cu-O Supercondutor”, Jpn. J. of Appl. Phts., 27 (1988) 1476-1479.
8. H. Endo, J. Tsuchiya, N. Kijima, A. Sumiyama, M. Mizuno and Y. Oguri, “Thermal Stability of the High Tc Supercondutor in the Bi-Sr-Ca-Cu-O System”, Jpn. J. of Appl. Phts., 27 (1988) 1906-1909.
Kozo Ishizaki and Koji Matsumaru
Department of Mechanical Engineering, Nagaoka University of Technology, Nagaoka, 940-2188, Japan
This paper was also published in print form in "Advances in Technology of Materials and Materials Processing", 13 (2011) 19-23.