This is a Case Study for microstructure property investigation of ZnO Ceramic Varistor using Scanning Electron Microscope – Electron Beam Induced Current (SEM-EBIC) and Secondary Ion Mass Spectroscopy (SIMS). This work was undertaken at the Department of Materials, Imperial College London.

1.0  Synopsis 

This case study is provided as an example of my abilities to manage and use characterisation techniques such as scanning electron microscopy (SEM), energy dispersive x-ray analysis (EDX), electron beam induced current (EBIC) and secondary ion mass spectroscopy (SIMS) to investigate the microscopic electrical behaviour and the compositional variation of additives in ZnO ceramic varistors. The terrace EBIC pattern enabled a direct observation of the microscopic electrical behaviour at the grain boundaries. Investigation of the oxygen diffusion coefficient at the surface and along the grain boundaries inside the sample was facilitated by a line scan mode on the Atomika 6500 ion microprobe (SIMS instrument). 

1.2.  Background

ZnO varistors are voltage dependent resistors, exhibiting strong non-linear electrical characteristics with fast response time (≈1O^-9s) and are suitable in high power applications (1). Varistors are used to safeguard electronic devices from high voltage transients (Fig.1).The microstructure of ZnO varistors consists of a randomly oriented array of grains decorated by secondary phases at the grain boundary and at triple points. Four types of microstructure have been reported by E. Olsson and G. L. Dunlop (2). The varistor phenomenon although largely attributed to the dopants decorated grain boundary interface (where dopants accumulate); it is also influenced by the presence of these additional phases. Models based on the Schottky barrier have been proposed to explain the varistor phenomenon (3). The quality of the varistor depends on the quality of these barriers corresponding to the grain/grain boundary interface. A group of symmetrical back to back Schottky barriers are modeled to represent the microscopic electrical behaviour of the ideal varistor. In practice, however, there are good and bad barrier grain boundary interfaces (3, 4). The oxygen concentration at the grain boundary interface is thought to influence the barrier height and is therefore critical for the varistor phenomenon (2,5).Since the discovery of the ZnO based varistor, several experimental techniques have been used to investigate the microstructure and the electrical properties attributed to the grain boundary interface. However, much about the local variations in the grain boundary properties still remains to be investigated. Bernds et al (6), (1984) reported using SEM-EBIC to investigate ZnO varistor ceramics. They used a lock-in-amplifier and an electron beam chopping facility to separate the electron beam induced current (EBIC) from the dc bias current and found that the EBIC signal showed a strong variation with bias voltage. Recently (1988) Gupta et al, (7) used SIMS to study the presence and distribution of Na in the microstructure of ZnO varistors. The presence of Na improved the stability of the ZnO varistor. Rossinelli et al, (8) used SIMS to verify the distribution of Bi at the gram boundaries. They found that Bi was quite uniformly distributed perhaps in a monolayer along the grain boundary, but in the direction perpendicular to the interface, the Bi signal decayed very rapidly.In this work it is intend to interpret the non-linear V:I characteristics of the ZnO varistor in terms of the microscopic electrical behaviour and compositional variations at the grain boundary interface. Some experimental results acquired from investigations, using a combination of techniques namely SEM-EDX, SEM-EBIC and SIMS are presented.

2.0    Technical content – RESULTS AND DISCUSSION 

2.1        Sample Samples used for SEM-EDX and SIMS analysis were synthesised using ZnO powder doped with 0.5 Mol % of each of the oxides of Sb, Mn, Bi, Cr, and Co. They were ball milled, calcined at 700°C for 5h., sieved, palletised in a steel die under a pressure of 75 MPa, for about 30 seconds and then sintered at 1250°C  for 2h, in a furnace programmed to ramp at 4°C/mm. The densities of the sintered compacts were in the range 96-98%. The sample used for EBIC investigations was a commercial ZnO varistor (60V), cut perpendicular to the two contacts. All the samples were polished using powders down to (0.25µm) diameter prior to analysis. The sample used for the SEM­EDX investigations was etched in perchloric acid in order to reveal phases at the triple points (Fig. 2a). 2.2 SEM-EDX SEM-EDX enabled characterisation of the microstructure and in particular phases in the etched specimen of ZnO varistor. The micrograph and the EDX spectra of the sample are presented in Figs.2a and 2b. Regions marked as la, 2, 3, 4 are pyrochlore phases rich in Bi + Sb and those marked as 1b and 5 are spinel, rich in Sb + Cr, as previously identified by XRD (9,10); These phases are also seen in Fig.3a, as a light (Bi rich) and slightly darker (Sb + Cr rich) features decorating the much darker ZnO grains in the microstructure. Phases of this types are known to influence the oxygen concentration at the grain boundary interface (5). 

2.3        SEM-EBIC In this work, EBIC was used to study the microscopic electrical behaviour of the ZnO varistor. The experimental arrangement is schematically shown in Fig.4. An electron beam current of lOOnA with 25KV acceleration voltage were found to be favorable experimental conditions for obtaining a good EBIC contrast. From SEM-EBIC studies, the terrace electrical pattern as seen in Fig.3b, appears to correspond to the grain boundaries of the ceramic decorated by secondary phases, seen in the compositional image in Fig.3a. This is perhaps an example of an array of electrically active grain boundaries in ZnO varistor. It is also seen that some grain boundaries do not display any significant terrace contrast and they are presumed to correspond to bad grain boundary junctions. The quality of a varistor would then depend on the ratio of these good/bad junctions in the ZnO ceramic system. The quality of these grain boundary junctions depends mainly on the procedures adopted in manufacturing these devices and in particular the control over the sintering and post sintering treatments. SEM-EBIC offers a useful method of characterisation of these devices. A typical voltage drop of about 3V is usually expected across a single grain boundary (3). In subsequent works single grain boundary interface was studied using a micro-manipulator and a line scan profiling (LSP) technique (11). 2.4 SIMS The SIMS technique was used to measure the compositional variation of the dopants in the ZnO varistor. Normal depth profiling with checkerboard gating, ion imaging, and line scan are amongst the analytical options available on our SIMS instrument (Atomika 6500 ion microprobe). Previously I had used both ion imaging and depth profiling to investigate the compositional variation of dopants (Bi, Sb, Co, Mn and Cr) in ZnO ceramic. From the results (11) it was found that dopants such as Co and Mn distributes uniformly and Cr, Sb and Bi segregate to form spinel and pyrochlore phases. In my subsequent SIMS work (11), I  investigated the spatial distribution of oxygen using a combination of imaging and line scan options.  Samples of pure ZnO ceramic and of the varistor composition described above were initially polished to 0.25µm. followed by heating to 1000°C for 16 h. in an ¹⁶0 environment (940 m Bar), and then subjected to an ¹⁸0 exchange for similar exposure times, temperatures and pressures. The ¹⁶0/¹⁸0 isotopic exchange was facilitated in a suitable rig (11). The ¹⁸0 isotope of oxygen has low natural abundance (0.2%) and it is used as a tracer to study the diffusion of oxygen. The raw data of ¹⁸0 diffusion in ZnO ceramic and the isotopic ratios of ¹⁸0 in pure and doped ZnO ceramic are presented in Figs.5 and 6. A diffusion tailing edge is revealed, diminishing right down to background values for pure ZnO but remaining just above the back ground (0.2%) and continuous across the sample of the varistor composition. A combination of both the solution to the diffusion equation and the tailing function has been fitted to the experimental data. From these data the diffusion coefficient (D) and the surface exchange coefficient (k) of a varistor sample have been estimated to be 0.79 x 10^-15 cm² sec^-1 and 0.14 x 10^-9 cm sec^-1 respectively.A SIMS line scan taken over the cross section of the ¹⁸0 exchanged samples shows the oxygen diffusion profile across the whole sample (Fig.7). From the numerical processing of this data a map showing the variation of the ¹⁸0 isotopic ratio in the base of the line scan crater (for the varistor sample) is shown in Fig.8. In the first instance, I speculate that this fast diffusion along the grain boundaries is to be due to the secondary phase network of mainly Bi₂0₃. However, it is also possible that this fast diffusion may be due to the porosity in the ceramic.  The presence of a fast diffusion path inside the sample was also reported by Fujitsu et al (5). This was investigated in our subsequent works (11) and I found that there were faster diffusion paths at the grain boundary interface in a doped ZnO ceramic varistor than in the case of the un-doped or pure ZnO ceramic.

3.0   CONCLUSIONS 

1      SEM-EDX confirms a uniform distribution of Mn and Co in the matrix of ZnO ceramic varistor. Dopants such as Bi, Sb, and Cr segregate to form spinel (Zn₇Sb₂0₁₂) and pyrochiore (Zn₂Bi₃Sb₃O₁₄) phases at the grain boundary regions.

2      SIMS investigations provide a complementary result to SEM-EDX and additionally, from the line scan analysis, it is speculated that the spatial distribution of oxygen is enhanced in the regions decorated by dopants at the grain boundary interface. In ZnO ceramic with the varistor composition, the oxygen diffusion coefficient is much greater then in the case of the undoped ZnO ceramic where there is no significant oxygen concentration in the bulk. 

3       SEM-EBIC micrographs feature a kind of terrace pattern which appears to correspond to an array of electrically active grain boundaries in ZnO varistor. Regions with no significant terrace contrast may be an indication of a “bad” junction. This technique enables a direct observation of the microscopic electrical behaviour at the grain boundaries.

Acknowledgements

I would like to thank the Department of Materials for the use of these experimental techniques. I am grateful to my project supervisors: Dr. D. S. McPhail and Prof B. C.H. Steele. I am grateful to Prof D.B. Holt, Prof J A Kilner, and Dr. R.J. Chater, for their useful comments in this work. I would also like to thank my industrial (NPL) supervisors Drs. M G. Gee and M Stewart for their support in this work. Finally, I would like to thank SERC and NPL for funding this project.

REFERENCES

1.        L. M. Levinson and H. R. Philipp, Zinc Oxide Varistors – A Review. Ceramic Bulletin, Vol.65 No.4 pp639 (1986).

2.        Olsson and G. L. Dunlop, characterisation of individual interfacial barriers in a ZnO Varistor material. Appl. Phys. Vol. 66 No.8 pp. 3666-75 (1989).

3.        Kazuo Eda, Conduction mechanism of non-Ohmic ZnO ceramics, J.Appl. Phys. Vol. 49 No.5 pp 2964-72 (1978).

4.        Richard Einzinger, Grain boundary phenomena in ZnO Varistors, Mater. Res. Soc. Proc. Edited by G. E. Pike, C. H. Seager, H. J. Leamy, Vol.5 pp. 333-341 (1982).

5.        S. Fujitsu, H Toyoda and H. Yanagdia, The Enhanced Diffusion of Oxygen in ZnO Varistor. Nippon-Seramikkusu-Kyokai-Gakujutsu-Ronbunshi Vol.96 No.2 pp.119-23 (1988).

6.        Bernds K.,Lohnert and E. Kubalek, SEM EBIC Investigation of ZnO varistor ceramics. Journal Dc Phys. Vol.2. No.45. pp.c2-861-64 (1984).

7         Gupta T. K. and Miller A.C., Improved Stability Of The ZnO Varistor via donor and acceptor doping at the grain boundary. J. Mater. Res.Vol. 3 No.4 pp. 745-754  (1988).

8.        Rossinelli M. Blatter and F. Greuter, Grain Boundary Properties of ZnO Varistor. Brown Boveri Research Center CH-5405 Baden, Switzerland, Brit. Ceram. Soc. (36) edited by B.C.H. Steele, pp.1-17 (1986).

9.        R. D. K. Pindoria, (Transfer report) Characterisation of ZnO ceramic varistor, Dept. of Materials, Imperial College, London (1991).

10.      M. G. Gee and M. Stewart, Some Observations on the Structure of ZnO Varistors. National Physical Laboratory, Teddington, Middlesex. Brit. Ceram. Proc. Vol. 42, pp 213 (1989).

11.      R. D. K. Pindoria, (PhD Thesis) Microstructure Property Investigation of ZnO ceramic varistor, Dept. of Materials, Imperial College, London (1995).

Ravji D Pindoria PhD DIC MSc BSc PGCE

Some useful links to my my knowledge based site and my my professional review report

Ref: Dr Dennis Kunkel