Production of ceramic turbine components by electrical discharge machining

1. Introduction

Advanced engineering ceramics are attracting more and more attentions in the last decades and employed as critical components in the modern mechanical systems because of their excellent mechanical, physical and chemical properties. Among those various ceramic materials (Al2O3, ZrO2, B4C, ...), Si3N4 has being regarded as one of the best ceramic material for structural application, owing to its low density, high hardness, high strength at high temperature, oxidation and thermal shock resistance. However, these promising properties also bring difficulties in machining and structuring this material, the wider applications in the industry thus are limited. In recent years, a successful approach by incorporating electrically conductive reinforcements such as TiN, TiC, TiCN, and TiB2 etc. into the silicon nitride matrix is developed [1]-[3]. Among all these mentioned secondary phase materials, the introduction of 30-40 vol. % TiN not just dramatically increases the electrical conductivity of the composite, but also enhances the mechanical properties, for instance fracture toughness, strength and wear resistance [4]. On the other hand, the composite is difficult to be machined efficiently by conventional manufacturing methods. Nevertheless, the significantly lowered electrical resistivity provides the possibility of structuring the composite by using other methods such as Electrical Discharge Machining (EDM). As it attracts more and more attention in the last few years, major ceramic suppliers and merchants have already commercialised this composite especially for the EDM application.

In this paper, the machining properties of Si3N4-TiN composite are evaluated via micro milling- and sinking-EDM. The miniaturized pulse energy input is the main differences comparing to macro-scale machining. Surface quality, microstructure, material removal behaviour regarding to the various condition are investigated and discussed. As an application example, the micro-manufacturing of the Si3N4-TiN composite into a miniature turbine impeller, which is a crucial component in a micro power generation system, with obtained the knowledge are demonstrated.

2. Experimental Investigation

2.1. Material properties and microstructure

The Si3N4-TiN ceramic composite employed in this research is obtained from a commercial ceramic supplier Saint-Gobain. Manufacturers grade is Kersit 601. The measured mechanical and physical properties of the composite are listed in Table 1.

2.2. Experimental Set-up

One of the tools applied in this research is a SARIX SX-100-HPM micro-EDM milling machine. The equipped relaxation type generator and high precision positioning system makes it dedicate for the micro-scale machining [8]. The short discharge duration time (350-400 ns) and low discharge current (< 0.5 A) guarantees the small energy input for each pulse. The usage of hydrocarbon oil as dielectric also assures the smaller sparking gap to attain the machining accuracy. The special designed spindle and clamping head allow the use of thin solid or tubular electrodes with diameters ranging from 40 m to 1.0 mm combined with external or internal flushing. The automatic electrode feeding system can compensate the tool length due to the wear during machining. However, the rotational speed is limited to 600 rpm.

Another tool is a die-sinking machine, Roboform 350γ from Charmilles. The advanced generator can generate various pulses from iso-energetic static to relaxation type. It also uses hydrocarbon oil as the dielectric.

Supplier Saint-Gobain
Grade Kersit 601
Chemical composition Si3N4 64 vol%
TiN 36 vol%
Binder Al2O3
Grain size (m) See Fig. 1
Density (ISO 3369) (g/cm3) 3.97
Hardness (ISO 3878), (kg/mm2) HV10 1508 33
HV30 1465 6
3-point bending strength (MPa) 979 120
Youngs modulus (GPa) 333 3
Fracture toughness (MPa.m1/2) 10 kg 8.7 0.7
30 kg 5.5 0.4
Resistivity (10-7 Ω.m) 160
Thermal conductivity (Wm-1K-1) 20C 28
800C 19
Table 1: Mechanical and Physical properties of Si3N4-TiN ceramic composite

a) magnification 2000x b) magnification 10000x
Fig. 1. Scanning microscopic pictures of the Si3N4-TiN composites.
Phases: Grey = TiN; Black = Si3N4; White = WC milling ball contamination.

2.3 Experimental results

It is known that the contradictory effects of the tool electrode wear with the material removal rate and the surface quality in the EDM process; and all are very much related to the given technology parameters. Thus it is impossible to study all the combination of parameters on the outcomes of machining abilities. This research is more focusing on achieving a good surface integrity of the ceramic composites in various machining configurations. A more detailed process investigation can be found in [9].

2.3.1 Micro-EDM milling

In the micro-EDM milling of Si3N4-TiN composite, tungsten carbide solid rod is used in the experiments. The relationship of the input actual machining parameters with the material removal rate, tool wear and the surface quality are shown in Table 2.

Machining regime Actual open gap voltage (V) Actual discharge current (A) Material removal rate (mm3/min) Tool wear ratio (%) Roughness Ra (m)
Rough -100 10 0.305 0.05 2.37
Semi-finishing -100 5 0.173 0.92 1.26
Finishing -70 0.5 - - 0.75
Table 2: Machining properties of micro-EDM milling with WC tool electrode

For the finishing regime, the material removal rate and tool wear ratio are difficult to be concluded because of the extremely slow process and unstable machining condition. As can be seen, the micro-EDM milling properties of Si3N4-TiN have the same variation trend as the machining of steel: lower pulse energy is necessary for achieving better surface quality, but not in favour of the machining speed and tool wear.

To further optimize the surface quality after the micro-EDM milling process, a series of experiments are conducted with gradually lowered energy input of the pulse. In Fig. 2 the relationship between the actual discharge parameters and the surface roughness Ra is plotted. Apparently the surface quality can only be optimized to a certain extent. The smoothest surface quality obtained during the tests is 0.74 m Ra. Further reduction of pulse energy cannot provide any improvement of the roughness, unlike the situation for micro-EDM of steel.

Accordingly the surface topography is examined. In Fig. 3 a), the topography of the Si3N4-TiN after the semi-finishing process, which has a roughness Ra of 1.65 m, is shown. There are no regularly formed craters visible like on a normal EDMed steel surface; on the contrary, a porous, sponge-like surface is revealed. The same phenomenon is also observed on the finest micro-EDMed surface, as illustrated in the cross-section view of Fig. 3 b. This porous layer has a thickness approximately 10 times the Ra value.

Fig. 2. Surface quality optimization in relation with the actual discharge parameters.
a) Topography with Ra 1.65 m b) cross-section with Ra 0.75 m
Fig. 3. SEM pictures of micro-EDM milled Si3N4-TiN surface.

2.3.2 Die-sinking EDM

Though the die-sinking EDM machine is not delicately designed for micro manufacturing purposes as aforementioned, it still has modules or technology settings which are specialized for micro-machining. Parameters investigated in the following experiments are set in this range. Copper infiltrated graphite (POCO EDM-C3) is applied as the tool electrode material. Two strategies are followed regarding EDM regimes: using relaxation type pulses for obtaining low energy input, and using short high-energy iso-pulses. The machining parameters and the performances are list in Table 3.

Machining regime Open voltage (V) Charging current (A) Charging time (s) Capacitance (nF) Material removal rate (mm3/min) Tool wear ratio (%) Roughness Ra (m)
Relaxation, roughing -200 6 6.4 67 13.8 3.29 2.91
Relaxation, finishing -120 1 25 1.0 0.032 25.5 1.54
Iso-pulse -80 12 1.6 - 1.79 22.4 0.93
Table 3: Machining properties of die-sinking EDM with copper infiltrated graphite tool electrode
Relaxation regimes

Compared to EDM milling, the roughing regime for die-sinking EDM with relaxation pulses offers higher machining speed but at the expense of a higher relative tool wear and higher surface roughness. As for the finishing regime, even with very low energy input of the pulse, the roughness is somewhat higher than resulting from the milling process in semi-finishing regime. Furthermore, it also suffers from a low material removal rate and elevated tool wear.

The SEM micrographs in Fig. 4 show the surface topography after die-sinking with different relaxation regimes. Similarly to EDM milling, a porous and foamy topography is revealed. Apparently there are no dramatic changes in microstructure with varied pulse energy. Moreover, no visible subsurface micro-crack is examined even at high magnification. SEM views.

a) Roughing regime resulting in Ra 2.91 m b) Finishing regime resulting in Ra 1.54 m
Fig. 4. SEM pictures of surface topography after die-sinking EDM with relaxation pulses
Iso-pulse regime

Fig. 5 shows the results for die-sinking EDM with short but powerful iso-pulses. No foamy or porous layer is present in contrast to previous results based on relaxation pulses. Unfortunately, cracks appear in the surface.

a) Material clearly removed by melting and evaporation b) Surface cracks (top view)
c) Surface without porous and foamy layer d) Surface cracks (cross-sectional view)
Fig. 5. SEM pictures of surface topography after die-sinking EDM with iso-pulse regime.

2.4 Discussion

As the experimental results show, continuously decreasing the pulse energy in relaxation regime is not improving the surface quality of Si3N4-TiN. These improvements are limited due to the intrinsic material removal mechanism. The material removal mechanism is not just melting and evaporation, but also the involvement of chemical reactions, as proposed in [7]: the decomposition of Si3N4 and TiN at elevated high temperature above 1700 C:

Si3N4 3 Si + 2 N2

2 TiN 2 Ti + N2

This decomposition reaction results in a high material removal and much lower tool wear compared to the machining of steel.

Unfortunately, this reaction generates enormous amount of gas nitrogen gas bubbles which prohibits the formation of an inerratic craters and leads to the creation of the voids, resulting in a foamy and porous top surface structure as shown in the SEM figures aforementioned. Furthermore, it can also be seen that the energy differences in the pulses cannot change the primary microstructure of the surface.

As a proof of the recognized material removal mechanisms, elemental analysis EDAX (Energy Dispersive X-ray Spectroscopy) of the ceramic matrix, specimens after micro-EDM milling and die-sinking EDM are conducted.

Table 4 lists the quantification of the detected elements. As expected, the content of N is dramatically decreased comparing to the ceramic matrix because of the decomposition. There is also a trace of carbon on both EDMed surfaces, which indicates that there might be a material transfer from the tool electrode to the workpiece surface. However it cannot be confirmed because the machining environment is in hydrocarbon oil.

Elements C N O Al Si Ti
Matrix - 24.19 5.00 1.69 39.79 29.33
EDM Milling 24.00 13.06 4.06 1.21 30.37 27.29
EDM Sinking 11.13 16.92 10.79 1.40 27.86 31.90
Table 4: EDAX quantification (wt%) of each element at Si3N4-TiN matrix and the surface after EDM

3. Turbine manufacturing

The goal of this manufacturing research is the manufacturing of ceramic turbine impellers for ultra micro gasturbines. The optimised turbine impeller has a mixed axial-radial design and 8 blades with three-dimensional geometry. Both micro-EDM milling and die-sinking are employed in the manufacturing of the prototype turbine impellers. The schematic views of the die- sinking process and the milling process are illustrated in Fig. 7 and Fig. 8, respectively.

Fig. 7a. A schematic view of the die-sinking machining process and an enlarged view of the tool electrode, as well as a turbine impeller.

Fig. 7b. Turbine with improved surface quality thanks to the use of the optimised iso energetic pulse regime.

Fig. 8. A close view of the micro-EDM milling process and a finished turbine impeller

In die-sinking process, the graphite electrode, which has a negative shape of a cavity, is produced on a 5-axis micro-milling machine (Kern MMP) with 3-axis machining configuration [10]. In total 10 electrodes are used for manufacturing a turbine. One of the electrodes and the finished product are shown in Fig. 7. Since the electrode milling and die-sinking process can run parallel, the total production time is approximately 15 hours.

A close view of the micro-EDM milling process for the turbine manufacturing is presented in Fig. 8. The milling of each cavity starts with 1.0 mm WC tool for pocketing and 0.7 m tool for wall finishing, with layer-by-layer milling process. And the layer thickness is 8 m and 3 m, respectively. Due to the low machining speed of finishing regime for this method, only roughing setting is applied. Even with relatively high material removal rate and no need for electrode preparation, it still takes 20 hours for machining one cavity. Thus total machining time for manufacturing a turbine impeller is about 160 hours.

4. Conclusion

Micro-scale electrical discharge machining of commercially available Si3N4-TiN conductive ceramic composites has been performed and the results reveal the attractive machinability of EDMing this hard, brittle material. The influences of pulse energy on the material removal rate, tool wear and the surface integrity are investigated.

With relaxation pulses, the improvement of the surface roughness is limited due to the chemical decomposition of the Si3N4 and TiN as a material removal mechanism. The surface also reveals porous, foamy structures.

With iso pulses, more regular craters are obtained and no foamy structures are observed. A moderately smooth surface quality (~ 1 m) can be easily achieved with rather high machining speed. However, tool wear when using iso pulses is high (~25%), comparable with the finishing regime using relaxation type pulses. As the iso pulses produce no porous foamy surface, and taking into account its competitive machining speed, it's a good replacement for the original finishing setting with relaxation pulses. Despite the better surface roughness, the flexural strength is not better than for relaxation type pulses. This reduced strength is caused by the microcracks due to the thermal impact formed at the surface. However, the attained strength is still sufficient for the application.

Further investigation on the formation regularly craters, influences of other technical parameters on the performances, the examination of chemical compound and internal/tensile stresses still need to be investigated.

Precision manufacturing of a miniature turbine with Si3N4-TiN ceramic composites by either die- sinking or micro-milling process has also been proved. Micro-milling EDM is obviously the slower process.


[1] Chu CY, Singh JP, Routbort JL, High-temperature failure mechanisms of hot-pressed Si3N4 and Si3N4/Si3N4-whisker-reinforced composites, Journal of the American Ceramic Society, vol. 76 (1993) pp 1349-1353.

[2] Xu HHK, Ostertag CP, Braun LM, Effects of fiber volume fraction on mechanical properties of SiCfiber/ Si3N4-matrix composites, Journal of the American Ceramic Society, vol. 77 (1994) pp 1897-1900.

[3] Shin DW, Tanaka H, Low-temperature processing of ceramic woven fabric/ceramic matrix composites, Journal of the American Ceramic Society, vol. 77 (1994) pp 97-104.

[4] Herrmann M, Balzer B, Schuberrt C, Hermel W, Densification, microstructure and properties of Si3N4-Ti(C,N) composites, Journal of the European Ceramic Society. Vol.12 (1993) pp 287-296.

[5] Liu CC, Huang JL, Effect of the electrical discharge machining on strength and reliability of TiN/ Si3N4 composites, Ceramics International, vol 29 (2003) pp 679-687.

[6] Liu CC, Microstructure and tool electrode erosion in EDMed of TiN/ Si3N4 composites, Materials Science and engineering A363 (2003) pp 221-227.

[7] Lauwers B, Kruth JP, Liu W, Schacht B, Bleys P, Investigation of the material removal mechanisms in EDM of composite ceramic materials, Journal of Materials Processing Technology, Vol 49 (2004) pp 347-352.

[8] Liu K, Ferraris F, Peirs J, Lauwers B, Reynaerts D, Process capabilities of micro-EDM and its application, Proceedings of the 3rd International Conference on Multi-Material Micro Manufacture, 3-5 October 2007, Borovets, Bulgaria, pp.267-270.

[9] Liu K, Ferraris E, Peirs J, Lauwers B, Reynaerts D, Process investigation of precision micromachining of Si3N4-TiN ceramic composites by electrical discharge machining (EDM), Proceedings of the 15th International Symposium on Electromachining (ISEM), 23-27 April 2007, Pittsburgh, PA, USA, pp 221-226.

[10] Ferraris E, Liu K, Peirs J, Bleys B, Reynaerts D, Production of a miniature Si3N4-TiN ceramic turbine impeller by die-sinking EDM, Technical Digest, The 7th International Workshop on Micro and Nanotechnology for Power Generation and Energy Conversion Applications, 28-29th November, 2007, Freiburg, Germany, pp 229-232.

Reply on content: Jan.Peirs at
Copyright 2008 Katholieke Universiteit Leuven, Department of Mechanical Engineering