MICRO ABRASIVE JET MACHINING OF CERAMICS M. WAKUDA Synergy Ceramics Laboratory, FCRA, Shidami, Moriyama-ku, Nagoya 463-8687, JAPAN; e-mail: wakuda@nirin.go.jp Y. YAMAUCHI and S. KANZAKI Synergy Materials Research Center, AIST, Shidami, Moriyama-ku, Nagoya 463-8687, JAPAN; SUMMARY Due to technical and economic advantages over existing micro machining technologies, abrasive jet machining (AJM) attracts much attention as a powerful method for dimpling of hard and brittle materials. This study investigated the performance of AJM process for various structural ceramics, and it has been found that both hardness and toughness of the materials are important factors affecting the machinability. The equation forming the relationship between the material properties and the material removal rate, established in the field of ceramic erosion, did not necessarily fit the AJM test results, because the machinability of the ceramic materials differed greatly depending on the employed jet particles. When the hardness of the abrasive is equivalent to, or lower than that of the ceramic material, the workpiece surface is roughened, but no dimpling takes place. Once the abrasive hardness exceeds that of the workpiece, smooth dimples can be generated mainly in a ductile manner, while brittle behaviour is also observed when machining low toughness ceramics with very hard abrasives.

Keywords: Abrasive jet machining, ceramic, erosion, machinability, material removal behaviour

1 INTRODUCTION Abrasive jet machining (AJM) is considered to be one of the most attractive techniques that can engrave precise dimples on the surface of hard and brittle materials [1, 2]. Although some practical uses of AJM have already demonstrated its high potential as a micro machining method capable of replacing other non-traditional processes, the detailed machining behaviour, for ceramics in particular, is still unknown. In general, AJM is categorised as blast finishing. The machining technique is however distinguished from traditional shot blasting in that it features a precision nozzle of less than 1 mm in diameter, through which a controlled mass of abrasive particles is continuously directed to the workpiece surface. As a consequence, AJM can meet requirements for patterning highly controlled micro dimples. Additionally, the machining action by AJM is shockless and any heat generated is dissipated by the enveloping gas stream. These factors provide a large advantage when compared with other micro machining methods such as ultrasonic machining and laser beam machining. From another point of view, AJM is a machining method positively utilising erosive wear behaviour, where fine hard particles attack the workpiece incessantly [3]. It is therefore attractive, not only for mechanical engineers but also tribologists and material engineers, to analyse the machining mechanism of the AJM process. However, due to the fact that the size of the particles employed in AJM is usually much smaller than that used in erosion tests, it is doubtful whether the established theories concerning erosion can be also applied. This study sets a target of clarifying the performance of AJM for various structural ceramics. Machinability is compared among four kinds of well-known ceramic materials. Microscopic observation provides a detailed

understanding of the material removal behaviour during AJM of ceramics. 2 EXPERIMENTAL PROCEDURE Machining experiments were carried out with a micro-blaster (MB2–ML–001, Sintobrator Ltd.) shown in Figure 1. This machine is capable of shooting fine abrasives along with a pressurised nitrogen gas stream through a small jet nozzle.

Jet nozzleSpecimen

Vacuum duct

XY stage

Figure 1: AJM experimental set-up

Abrasive type Abrasive size Jet pressure Jet distance Mass flow rate

WA, GC, SD 15–25 µm 0.30 MPa 0.5 mm 2 g/min

Table 1: Experimental AJM conditions

Principal machining conditions are listed in Table 1. The abrasive grit was mixed with the gas stream ahead of the nozzle, and the mass flow rate was kept constant throughout the machining process. The jet nozzle was made of tungsten carbide for wear resistance and had a bore diameter of 0.6 mm.

The employed cutting media were aluminium oxide (WA) and silicon carbide (GC), which are the most common abrasives in the application of the AJM process. Synthetic diamond (SD) was also used to investigate the effect of the abrasive hardness on the machinability. Although the use of the expensive super-abrasive is unpractical in the AJM process where jet particles are not available for recycling, its high hardness was very attractive in machining of hard ceramic materials from an empirical aspect. The mesh size of the utilised abrasives was #800, corresponding to the grain size range 15–25 µm, selected after the preliminary tests. Too rough abrasives resulted in blocking of the nozzle, whilst finer abrasives were difficult to draw through the nozzle system. The materials used in the tests were four kinds of well-known structural ceramics, Si3N4, SiC, Al2O3, and ZrO2. They were commercial samples whose mechani-cal properties are shown in Table 2. An important thing to be emphasised is that the hardness increased in the order ZrO2, Si3N4, Al2O3, SiC and the fracture toughness increased in the order SiC, Al2O3, ZrO2, Si3N4. In short, these two parameters of the mechanical properties had almost inverse correlation.

3 RESULTS AND DISCUSSION

3.1 Dimpling of Various Ceramics by AJM

The appearance of the dimples following abrasive jet machining of 10 seconds is shown in Figure 2. It was recognised that the properties of the dimples, not only in terms of the removed volume, but also the roughness of the struck face, differed greatly depending on the combination of jet particles and workpiece material. When WA abrasive was used for such hard ceramics as Al2O3 and SiC in particular, slight surface roughening occurred, but no dimpling. For the other machining set, smooth dimples with an inverted dome shape could be generated, and few obvious defects such as cracks and chipping were observed on the new-born faces.

For the purpose of evaluating the machining efficiency quantitatively, the material removal rate was calculated for each machining set and it is shown in Figure 3. The phenomenon common to all the ceramics is that the removal rate tended to increase with the increase of abrasive hardness from WA to SD. However, the trend was different depending on the workpiece ceramics. The machining efficiency of SiC dramatically increased with the hardness of the jet abrasive, while that of ZrO2 exhibited little dependence on the abrasive hardness.

ZrO2 Si3N4 Al2O3 SiC Density (g/cm3) Young's modulus (MPa) Vickers hardness HV (GPa) Fracture toughness KIC (MPa·m1/2) Flexural strength (MPa)

6.05 210 13.2 7.0

1200

3.2 290 14.2 7.5

1000

3.9 390 15.3 4.2 360

3.1 390 22.1 2.5 470

Table 2: Material properties of the tested ceramic samples

ZrO2 Si3N4 Al2O3 SiC

WA

GC

SD

Workpiece material

Jet a

bras

ives

200 µm

Figure 2: Appearance of the AJM face for various machining set

ZrO2 Si3N4 Al2O3 SiCWA

GCSD

0

0.005

0.010

0.015

0.020M

ater

ial r

emov

al ra

te

mm

3 /s

Workpiece Jet p

articl

e

Figure 3: AJM machinability of various ceramics

Interestingly, it was found that when using WA abra-sive, the material removal rate of the workpiece showed an inverse correlation with their hardness. Material removal rate decreased in the order ZrO2, Si3N4, Al2O3, SiC. However, when the hardest abrasive (SD) was used, the order of the material removal rate correlated with their fracture toughness. The material removal rates when using GC abrasive were in between those for WA and SD, and it is thought that the influence of both hardness and toughness was combined.

With regard to the erosive behaviour of ceramics, there are papers, which have attempted to develop a relationship between the erosion rate and these two parameters [4–6]. The data are fitted to the following formula:

∆E ∝ H α · KC β (1)

where ∆E is the erosion rate, H is the hardness, KC is the toughness, and α and β are non-dimensional constants.

As an example, the best-known equation was given by Evans [4] as follows:

∆E ∝ H –1/4 · KC –4/3 (2)

The equation was developed through theoretical analysis of particle impact on ceramics, and it was also verified that it correlated with the empirical data [7].

In the case of the above-mentioned AJM test results, however, the individual effects of hardness and toughness on the material removal rate varied with the impacting abrasive. In other words, the exponents of Equation (1) can not be thoroughly determined so as to fit all the machining data. This is attributed to the fact that the machining mechanism changes when subjected to abrasive jet in a variety of tool-workpiece combinations.

3.2 Material Removal Behaviour in AJM Process

Before observing the dimpled faces in detail, material response to an abrasive impact was modelled with an indentation test. Figure 4 shows the surface appearance following the indentation by a diamond spherical indenter (spherical radius: 0.2 mm, load: 98 N, loading duration: 15 s). Two different modes of material response were found: ductile and brittle modes. These behaviours are important features when discussing a machining mechanism [8]. While the material behaviour in the former mode is comparable to that of metals, in the latter mode the material is fractured by crack propagation. In the figure, for the materials with low hardness and high toughness, ductile behaviour was predominant. For the hard and less tough materials, on the contrary, brittle behaviour became major and it resulted in unexpected large-scale fracture.

Al2O3 SiC

10 µm

ZrO2 Si3N4

Figure 4: Ceramic surface following indentation test

Bearing these two modes in mind, microscopic observation with higher magnification was executed for the AJM dimples. As the surface appearance of the Al2O3 sample varied most significantly with the employed abrasive, the following discussion uses the photomicrographs of the alumina sample (Figure 5).

First, the high-magnification view of the surface machined with WA, Fig. 5(a), consisted of craters caused by plastic deformation, corresponding to the jet abrasive size, and flaking surrounding the plastic zone. The mixture of these two phenomena formed a very rough surface. It is thought that due to the lack of the abrasive hardness compared to the workpiece material, most of the abrasives were able to produce little plastic deformation on the sample, but micro flaking occurred by grain boundary cracking. This is the reason for the

10 µm

(a) (b) (c)plastic zone

Figure 5: High magnification of the Al2O3 faces machined with (a) WA, (b) GC, and (c) SD abrasives

occurrence of surface roughening. Thus, when WA grit is employed, the machinability is apt to be dominated by the material hardness, and the relative hardness of the abrasive with respect to that of the workpiece, is an important factor as to whether dimpling occurs or not.

Second, for the combination of GC abrasive and the alumina sample, shown in Fig. 5(b), an extremely smooth face was obtained. The jet abrasive was now hard enough to engrave the workpiece, and the surface was therefore deformed plastically via ductile regime machining, in agreement with machining of metals. Almost all the machining results for ZrO2 and Si3N4 resembled this ductile appearance.

When using SD abrasive, on the other hand, the surface shown in Fig. 5(c) was not as smooth as that machined with GC. The main feature of the material response might still be plastic deformation, but this time it was accompanied by crushing triggered by grain boundary cracking. It is supposed that a remarkable increase of the material removal rate resulted from the large-scale fragmentation occurring in a brittle manner, and the trace of the brittle fracture remained on the final ductile-dominant face. In addition, the dimpled faces of SiC abraded by GC and SD also showed brittle appearance, which caused the dramatic increase of the material removal rate. It is considered that the exposure of brittle behaviour when using SD abrasive is the main reason why the fracture toughness became to dominate the machinability. 4 CONCLUSIONS The machinability of various structural ceramics by the micro AJM process, in other words the erosive wear behaviour by fine-grained abrasives, was evaluated.

When the abrasive hardness is equivalent to, or lower than that of the ceramic material, the workpiece surface is roughened, but no dimpling takes place. It conforms to the general law of machining processes. Once the abrasive hardness exceeds that of the workpiece and the difference in their hardness becomes sufficiently large, the material toughness tends to dominate the machinability. It is not possible to generalise the

obtained data in the form of the equation established in the study on ceramic erosion, because the machinability varies greatly with the impacting abrasive.

According to the observation at higher magnification, most of the dimpled faces, for high toughness ceramics, in particular, exhibit ductile appearance. Brittle behaviour appears in the machining of low toughness ceramics, which is the reason for the dramatic increase of the material removal rate. 5 ACKNOWLEDGEMENT This work has been supported by METI, Japan, as part of the Synergy Ceramics Project. Part of the work has been supported by NEDO. The authors are members of the Joint Research Consortium of Synergy Ceramics. 6 REFERENCES [1] Shukla, R. B.: Abrasive Jet Machining. Proc. Abra. Eng. Soc. Conf., 23 (1985), 91–100 [2] Herbert, D.: Blast Finishing. Metal Finishing, 97 (1999) 1, 93–100 [3] Zhang, Y.: Cheng, Y. B.: Lathabai, S.: Erosion of Alumina Ceramics by Air- and Water-Suspended Garnet Particles. Wear, 240 (2000), 40–51 [4] Evans, A. G.: Gulden, M. E.: Rosenblatt, M.: Impact Damage in Brittle Materials in the Elastic-Plastic Response Regime. Proc. R. Soc. London, Ser. A. 361 (1978), 343–365 [5] Ruff, A. W.: Wiederhorn, S. W.: Erosion by Solid Particle Impact. Treat. Mater. Sci. Tech., 16 (1979), 69–126 [6] Wiederhorn, S. M.: Hockey, B. J.: Effect of Material Parameters on the Erosion Resistance of Brittle Materials. J. Mater. Sci., 18 (1983), 766–780 [7] Gulden, M. E.: Correlation of Experimental Erosion Data with Elastic-Plastic Impact Models. J. Am. Ceram. Soc., 64 (1981) 3, C59–60 [8] Trumpold, H.: Hattori, M.: Tsutsumi, C.: Melzer, C.: Grinding Mode Identification by Means of Surface Characterization. Ann. CIRP, 43 (1994) 1, 479–482