.I. Phys. Chew. SolLir Vol. 54, No. 8, pp. 951-954. 1993
oozz-3697/93 1.00 + 0.00 8 1993 FWgamon Press Ltd
Printed in Great Britain.
CHARACTERIZATION OF ZIRCONIA-TOUGHENED ALUMINA (ZTA) CERAMICS BY SEM-CATHODOLUMINESCENCE J. LLOPIS and S. E. PAJE Departamento de Fisica de Materiales, Facultad de Ciencias Fisicas, Universidad Complutense, 28040 Madrid, Spain (Received 2 December
1992; accepted 4 March 1993)
Ab&ract4thodoluminescence (CL) studies in a scanning electron microscope of zirconia-toughened alumina (ZTA) with three different zirconia contents are described. The CL spectra show a broad violet emission centered at about 3.1 eV (400 nm) and a large and complex blue-green band centered near 2.6 eV (475 nm). The former arises from alumina, whereas the latter is associated with the zirconia component. A correlation between the luminescence and the nature of monoclinic to tetragonal zirconia is found. The results are discussed on the basis of oxygen vacancies and complex defects. The effects of toughening mechanisms on the luminescence of the ZTA samples are also considered.
Keywords: ZrO,, cathodoluminescence, toughness, ceramics.
INTRODUCTION The growing interest in zirconia-toughened alumina (ZTA) ceramics is chiefly due to the remarkable improve in the mechanical and some of the physical properties with respect to pure alumina ceramics [l-3]. This has enabled the number of potential uses of alumina ceramics to be extended considerably. The improved mechanical properties of ZTA are primarily related to the stress-induced tetragonal to monoclinic zirconia phase transition. Thus, information about the fractions of metastable tetragonal and monoclinic phases and the factors which may influence stabilization as well as their spatial distribution in the alumina matrix are imperative for a good characterization. Previous studies on the luminescence of zirconia-based ceramics have shown that luminescence techniques can be used for structure characterization [4-q. The present studies were partially motivated by evidence that monoclinic transformed areas appear strongly luminescent but apparently have the same spectrum. The latter fact presents an added difficulty to the usual small size of ZrOz particles in distinguishing the different polymorphs present in the ceramics by CL-SEM (resolution power 1 pm). In addition, the detailed factors controlling the luminescence in ZrO, are not yet fully understood; in particular the role played by the polytypes. For this reason, a systematic study of the crystal structures of different polymorphs in relation to zirconia luminescence is in progress. The current study is the second part of a wider
investigation designed to characterize zirconia-toughened alumina ceramics by luminescence techniques. In the first part photoluminescence (PL) was used for characterization of ZTA samples whereas in this part cathodoluminescence (CL) in a scanning electron microscope (SEM) is utilized as the main procedure.
EXPERIMENTAL PROCEDURE Three ZTA samples previously investigated by PL [7j and labelled Al6Z, A9Z and A5Z were used. The samples contained 13.4, 7.4 and 3.9 wt% of Zr02, with fractions of the metastable tetragonal phase of 7.9, 5.9 and 3.0 wt%, respectively. The impurity content, the value of some physical properties and the average sizes for the alumina and zirconia particles of 1.4 and 0.6 pm respectively, were previously reported in [7,8]. The samples were indented with a load of 200 g using a Vickers microhardness attachment to a Zeiss optical microscope. Cathodoluminescence experiments were carried out with a scanning electron microscope (Cambridge S4-IO) operating at 20 kV with beam current over the range from 10e9 to 10m6A. For spectral analyses a light pipe which feeds the light into a 0.25 Ebert Jarrel-Ash monochromator was attached to the microscope and a photomultiplier tube 9558 B was used. The response of the system as a function of the wavelength, ranging from 380 to 650 nm, was calibrated against a standard lamp. Cathodoluminescence images and CL spectra from the three different ZTA samples, one standard of a pure alumina cer951
J. LLOPISand S. E. PME
amic and another of a magnesia-stablized zirconia single crystal, were analyzed. To study the spatial distribution of CL the light pipe and monochromator were replaced by a set of interference filters centered in the range 410-590 nm (3.c2.1 eV). All the samples were coated with a transparent conducting film of graphite and no spectral correction for the absorption was made. For the spectral analyses we used a personal computer with software based on the Marquardt-Levenberg algorithm.
PHOTON ENERGY (eV)
RESULTS AND DISCUSSION Figure 1 shows the normalized CL spectra of the samples labelled Al6Z, A9Z and A5Z. These spectra exhibited a main broad band peaked at about 3.1 eV (400 nm) with a prominent shoulder at about 2.6 eV (475 mn). The shoulder appeared better defined in the CL spectrum of the A9Z sample than in A162 and ASZ samples. In contrast with the PL emission spectra, previously described, the detailed shape of the CL spectra varies markedly depending on the zirconia content of the samples. The CL spectra were decomposed satisfactorily into Lorentzian and Gaussian components as shown in Fig. 1. Comparing the CL spectra with the luminescence from pure alumina and zirconia reported in the literature [5-7, 9-141, we are tempted to ascribe the origin of the 3.1 eV-band to the alumina fraction of the samples and the 2.6eV-band to the zirconia fraction. The former is quite similar to the luminescence band attributable to F centers (oxygen vacancy occupied by two electrons) in pure alumina [g-12]. On the other hand, the luminescence of zirconia usually appears in the blue-green region of the spectrum. The nature and detailed mechanism of this luminescence are not clear yet and have been a source of controversy. However, the results of recent studies strongly point towards F,-type (F-center with an impurity in the nearest-
PHOTON ENERGY (eV) Fig. 1. Decomposition of normalized CL spectra into Lorentzian and Gaussian peaks, dashed lines, for: (a) A16Z; (b) A92 and (c) A5Z samples, respectively. Full lines are the fitted curves.
Fig. 2. Normalized CL spectra for: (a) magnesia ZrO, single crystal standard and (b) alumina ceramic standard. The spectra are fitted to two Gaussian and a Lorentzian functions, respectively, shown by the full lines.
neighbor cation shell) or similar centers, favored by the tetragonal to monoclinic transformation, being involved in the zirconia luminescence . In order to verify the origin of the two main bands observed in the ZTA samples, CL spectra from standards of pure alumina and magnesia-zirconia, respectively, were recorded. The latter was selected because it displays a quite similar excitation and emission PL spectrum to those of the ZTA samples. Figure 2 shows the normalized CL spectra of the magnesia-stabilized zirconia single crystal and the pure alumina ceramic. Both spectra consist of structureless broad bands. The latter, with a full-width at half-maximum (FWHM) close to 0.49 eV and peaked at about 3.1 eV (400 nm), is quite similar to the Lorentzian band reported above for the ZTA samples. On the other hand, the CL band of magnesia-stabilized zirconia sample showed a FWHM of 0.72 eV with a maximum at about 2.48 eV (500 nm) and a weak shoulder in the range 2.17-2.05 eV (570-600 nm), in good agreement with previous observations [5-7, 13, 141.In this case, the CL spectrum can be decomposed into two Gaussian-shaped bands as shown in Fig. 2a. A comparison of the dominant band with the Gaussian of the ZTA samples showed that there was a 20-30 nm shift towards short wavelengths in the peak positions and slight variations in the half-width (FWHM) values. These apparent disparities in shape compared to previous observations may be caused by an absorption band in the blue-green region during SEM observations, since our CL spectra were not corrected for autoabsorption. In fact, after SEM examination all the ZTA samples exhibited a brownish coloration which is related to a 2.58eV (480nm) absorption band [ 15-171. A similar shift to short wavelength was found in the PL experiments . Taking into account the results mentioned above it is clear that the violet luminescence comes
of ZTA ceramics by SEM-CL
m/t Fig. 3. Ratio of relative intensities of the blue-green band, ZpL/ZcL,against the monoclinic/tetragonal fraction of the ZTA samples. The solid line through the experimental data represents a least-squares fit to y = 0.045 + 3.6x.
predominatly from the alumina grains and the blue-green luminescence essentially from the zirconia component of the ZTA samples. However, no immediate relationship between the intensities of the CL bands and the ZrO,, content was found. Thus, the A9Z sample with an intermediate ZrOz content showed the most intense blue-green band. An attempt to correlate the zirconia content with the luminescence of ZTA samples was made. For this purpose we used the CL results reported herein, in addition to prior photoluminescence data already reported in Ref. 7. Figure 3 shows a linear dependence between the ratio of the relative intensities, ZPLI&L 9 of the blue-green band and the m/t fraction of the ZTA samples. ZpL and ZcL are the relative intensities of the blue-green band upon exciting with 4.25 eV (292 mn) light and the electron beam of SEM, respectively. The experimental point near the coordinate origin in Fig. 3 was determined from the ratio at 2.48 eV (500 nm) between the PL and CL intensity in the spectra from the alumina standard. The results are consistent with the presence of competitive mechanical mechanisms affecting the luminescence response of the sample. On the one hand, there may be transformation t-m mechanisms in which radiative
recombination processes are favored (e.g. presence of radiative F-type and/or some complex centers), while on the other, there may be mechanisms with a detrimental effect on the luminescence which increase the probability of nonradiative recombination processes, such as internal damage. In this way, microcrack toughening would contribute to luminescence quenching while transformation mechanisms would result in an increase in luminescence. As pointed out in [7j, a combination of the two toughening mechanisms is likely to be present in the ZTA samples studied here. Panchromatic (all photon energies) and monocromatic (3.1 and 2.48 eV) CL images of the samples did not show a uniform emission and look similar to each other. Thus, the strongly luminescent areas observed for 2.48 eV-images also appear as bright luminescent areas in the 3.1 eV-images but with lower contrast, as shown in Fig. 4. Unfortunately, it was not possible to resolve individual grains in the CL images due to the small grain size of the alumina and zirconia crystals, near to the resolution power (N 1 pm) of the CLSEM. Occasional large and strongly luminescent grains, some pm in size, were observed on the surface of the A16Z samples. Indentation prints appeared in the CL images as no-luminescent areas for all the samples, while a slight increase in the CL intensity around indentations was observed. Both features are in good agreement with previous observations [4-6] and also consistent with a quenching of CL by mechanical damage and an increase in CL favored by the t-m transformation. According to the results mentioned above, it seems possible that the bright luminescent areas in the CL images are connected to nonstoichiometric areas created by the oxygen deficiency in the samples. In summary, luminescence, and in particular the CL technique, provide us with a method for studying the characteristics of ZTA ceramics, which include information about the monoclinic to tetragonal ratio and oxygen deficient areas present in the samples.
Fig. 4. SEM images from the A9Z sample: (a) emissive mode; (b), (c) and (d) CL mode with ail wavelengths (panchromatic image), 2.48 eV (500 run) and 3.1 eV (400nm) monochromatic images, respectively.
J. LU)PIS and S. E. PAJE
Toughening mechanisms the CL spectra.
seem to affect the shape of
Acknowledgements-The authors would like A. Ibarra and R. Vila for providing the ZTA finanical support of Direcciirn General de Cientifica y T&mica, DGICYT, project No. also acknowledged.
to thank Drs samples. The Investigation PB90-0250 is
REFERENCES 1. Claussen N., Ruhle M. and Hever A. H., Science and Technology of Zirconia II, Series Advances in Ceramics, Vol. 12. Am. Cer. Sot.. Columbus. Ohio (1984). 2. Butler E. P., Mater. SC;. Technol. i, 417 (i985j. 3. Becher P. F., Acta Metall. 34, 1885 (1986). 4. Czemuszka J. T. and Page T. F., J. Amer. Ceram. Sot. 68, 196 (1985). 5. Rincbn J. Ma, Femsindez P. and Llopis J., Appl. Phys. A 44, 299 (1987).
6. Llopis J., Phys. Star. Sol. (a) 119, 661 (1990). 7. Llopis J., Vila R. and Ibarra A., J. Phys. Chem. Solia!~ 52, 903 (1991). 8. Molla J., Ibarra A., Frost H. M. III, Clinard F. W. Jr., Kennedy J. C. III and Jimenez de Castro M.. J. NIX/. Mater. 179-181, 375 (1991). 9. Levy P. W., Phjs. Re;. lti, 1226 (1961). 10. Evans B. D. and Stanelbrock M.. Phvs. Rev. B 18.7089 11. Draeger B. G. and Summers G. P., Phys. Rev. B 19, 1172 (1979). 12. Lee K. H. and Crawford J. H., Phys. Rev. B 19, 3217 (1979). 13. Sarver J. F., J. electrochem. Sot. 113, 124 (1966). 14. Paje S. E. and Llopis J., Appl. Phys. A 55, 523 (1992). 15. Wright D. A., Thorp J. S., Aypar A. and Buckley H. P., J. Mater. Sci. 8, 1401 (1973). 16. PaiVemeker V. R., Patelin A. N., Crowne F. J. and Nagle D. C., Phys. Rat. 40, 8555 (1989). 17. Ben-Michael R., Tannhauser D. S. and Genossar J., Phys. Rev. B 43, 7395 (1991).