Nano grain sized zirconia–silica glass ceramics for dental applications

Available online at www.sciencedirect.com

Journal of the European Ceramic Society 32 (2012) 4105–4110

Nano grain sized zirconia–silica glass ceramics for dental applications Cecilia Persson ∗ , Erik Unosson, Ingrid Ajaxon, Johanna Engstrand, Håkan Engqvist, Wei Xia Division of Applied Materials Science, Department of Engineering Sciences, Uppsala University, Uppsala, Sweden Received 4 January 2012; received in revised form 31 May 2012; accepted 23 June 2012 Available online 26 July 2012

Abstract Glass ceramics based on lithium disilicates are commonly used in dental veneers and crowns. Alternative materials with improved mechanical properties may be of interest for more demanding applications, e.g. bridgeworks. In this study, a sol–gel method was optimized to produce nano grain-sized zirconia–silica glass ceramics with properties adequate for dental applications. The material properties were compared to those of IPS e.max® CAD, a commercially available lithium disilicate. The zirconia–silica glass ceramic was found to be translucent, with a transmittance of over 70%, and possessed excellent corrosion resistance. It also presented a somewhat lower elastic modulus but higher hardness than the lithium disilicate, and with the proper heat treatment a higher fracture toughness was achieved for the zirconia–silica glass ceramic. In conclusion, the material produced in this study showed promising results for use in dental applications, but the production method is sensitive and large specimen sizes may be difficult to achieve. © 2012 Elsevier Ltd. All rights reserved. Keywords: Sol–gel processes; Mechanical properties; Glass ceramics; Silica; Zirconia

1. Introduction Ceramics and glass ceramics are commonly used as dental materials due to their adequate mechanical properties and appealing aesthetics. The all-ceramics such as zirconia and alumina have the advantage of good mechanical properties, but are opaque and thus less aesthetically pleasing and more difficult to adapt to the colour of the surrounding teeth than glass ceramics such as lithium disilicates, which are translucent. However, the glass ceramics generally have a fracture toughness and flexural strength less than half that of the allceramics,1 which may be an issue for the mechanical resistance of e.g. longer bridgeworks.2 A translucent glass-ceramic with improved mechanical properties would therefore be interesting for dental restoration applications. A potentially stronger alternative to the lithium disilicates may be a glass-ceramic in the nano-grain sized zirconia–silica system. Nogami and Tomozawa3 reported on a low-cost sol–gel

∗ Corresponding author at: Division for Applied Materials Science, Department of Engineering Sciences, Uppsala University, Box 534, 751 21 Uppsala, Sweden. Tel.: +46 18 471 79 61; fax: +46 18 471 35 72. E-mail address: [email protected] (C. Persson).

0955-2219/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jeurceramsoc.2012.06.028

technique to produce translucent, high toughness glass ceramics in the ZrO2 –SiO2 system, by sintering directly the xerogel (as opposed to√powder processing). A fracture toughness of up √to 4.76 MPa m was reported, in comparison to the 4.15 MPa m reported for the lithium disilicate IPS e.max® Press using a similar method of evaluation.4 The high mechanical properties were attributed to the transformation toughening mechanism of tetragonal–monoclinic zirconia, as stable tetragonal zirconia particles were achieved through this sol–gel technique, without the addition of stabilizers such as yttria. However, this method for producing ZrO2 –SiO2 glass ceramics has not yet been evaluated for dental applications. Also, whereas the hardness of these systems has been evaluated for compositions of 5–50% ZrO2 (6.4–7.3 GPa),5 the fracture toughness has only been evaluated for 60% ZrO2 .3 No data on the elastic modulus is available, nor the corrosion resistance, which is important for dental applications.6 Furthermore, the translucency of these materials has been illustrated,3,7 but not measured. The aim of the present study was to optimize the manufacturing process and the material composition of the ZrO2 –SiO2 system in order to achieve a translucent, high toughness material suitable for dental applications. To this end, glass ceramics containing 30–40% ZrO2 were produced and the translucency,

4106

C. Persson et al. / Journal of the European Ceramic Society 32 (2012) 4105–4110

corrosion resistance, crystallinity, elastic modulus, hardness and fracture toughness were evaluated.

higher interest, as well as the increased difficulties in producing samples with a higher ZrO2 content (due to increased risk for self condensation and precipitation during preparation).

2. Materials and methods 2.2. Methods 2.1. Material synthesis ZrO2 –SiO2 glass ceramics of 30, 35 and 40 mol% ZrO2 were produced, using a sol–gel method adapted from Nogami and Tomozawa.3 A drying control additive was not reported as being used in their study, but was found useful in the present study for reducing the number of cracks in the specimens. Sol–gels were produced using the alkoxide precursors tetraethyl orthosilicate (TEOS) and 70 wt.% tetrapropyl zirconate (TPZ) in 1-propanol (all chemicals used in this work were acquired from Sigma–Aldrich, St Louis, MO, USA). Synthesis was initiated by mixing ethanol (EtOH, >95%), aqueous hydrochloric acid (HCl) and the drying control additive dimethylformamide (DMF) in a 50 ml round bottom flask, followed by the addition of TEOS under continuous stirring. The resulting molar ratio of the solution was 1:1:1:1 – TEOS:DMF:EtOH:H2 O. The partially hydrolysed TEOS was then magnetically stirred for 3 h in order to obtain a clear sol and ensure its homogeneity. The desired amount of TPZ was then added slowly using a micropipette and magnetic stirring of the solution was continued overnight. As EtOH is volatile, the sol was always kept covered during mixing in order to minimize any changes in concentration due to evaporation. Depending on the amount of zirconium alkoxide precursor added to the solution, aqueous (0.15 M, 0.40 M or 12.18 M) HCl was then added drop by drop to initiate the final hydrolysis and polymerization of a monolithic gel. After the final synthesis step, the solution was divided and transferred to Teflon® moulds of 25 mm diameter, which were sealed with polymer film for controlled evaporation. The sols were left to gel and age until approximately 40% shrinkage was observed and a stiff gel was formed. The highest yield (less amount of cracked samples and samples that were easily detached from the mould) was obtained when the polymer film was perforated after one week. The subsequent ageing time was then approximately three weeks. Note that the use of a hydrophobic container is essential to reduce capillary stresses otherwise leading to crack formation during drying. The material at this point is to be considered as fragile and any premature heat treatment, as well as any form of physical disturbance should be avoided. After formation of a stiff xerogel, samples were moved to an oven and kept at 100 ◦ C in an atmosphere of 100% relative humidity for 5 h. The temperature was then raised to 150 ◦ C and held for 15 h. The subsequent heat treatment process was initiated with a calcination plateau at 800 ◦ C with a holding time of one hour. Sintering of the samples was then carried out at 1100 ◦ C, with holding times of 10 or 15 h. The ramping rate was 20–30 ◦ C/h. Analysis was performed on samples containing 30, 35 and 40% ZrO2 , sintered at 1100 ◦ C for 10 or 15 h. However, samples containing 40% ZrO2 were sintered for 10 h only, as the results indicated that the samples containing less ZrO2 were of

The synthesized materials were evaluated using translucency measurements, X-ray diffraction, nano- and microindentations and corrosion resistance measurements. 2.2.1. Translucency Transmittance of samples was measured in the visible spectrum (380–750 nm) using a Lambda 900 spectrometer (PerkinElmer, Waltham, MA, USA) equipped with an integrating sphere detector, coated with Spectralon® (Labsphere Inc., North Sutton, NH, USA). Transmittance is defined as the ratio of the intensity of transmitted light, I, and the intensity of the incoming light, I0 : T =

I I0

(1)

Measurements were done on samples with a thickness of approximately 1 mm. A sample containing 35% ZrO2 was evaluated as well as a reference specimen of IPS e.max® CAD (Ivoclar Vivadent AG, Schaan, Liechtenstein). 2.2.2. Crystallinity The crystallinity of the samples was analysed by X-ray diffraction analysis using a Siemens Diffractometer D5000 (Siemens AG, Munich, Germany). Parallel beam diffraction was used for bulk scans and grazing incidence diffraction (GIXD) was used for surface sensitive scans. The analysis was performed over the 2θ range of 10–80◦ and at a step size of 0.05 and scanning speed of 5 s/step. For the GIXD an incidence angle of 2◦ was used. The particle size was calculated using Scherrer’s formula: t=

K·λ B · cos θB

(2)

where t, mean size of crystallite; K, constant dependent on crystallite shape (0.89); λ, X-ray wavelength; B, FWHM (full width at half max); θ B , Bragg angle. The specific peak at around 30◦ was selected to calculate the crystallite size of tetragonal ZrO2 . To evaluate the effect of grinding on the crystallinity of the samples a sample was ground for approximately 10 min and the surface of the sample was subsequently studied by GIXD. This study was prompted by the possible stress-induced transformation of the tetragonal ZrO2 , as the high fracture toughness is thought to be due to transformation of the tetragonal phase into the monoclinic phase. Nogami and Tomozawa3 did not conduct polishing or grinding after heat treatment in order to avoid this transformation. 2.2.3. Mechanical properties Preparation of samples for nano- and microindentation included mounting in thermoplastic resin, wet grinding and

C. Persson et al. / Journal of the European Ceramic Society 32 (2012) 4105–4110

4107

polishing. Grit papers from 320 to 1200 were used and polishing was done using diamond pastes from 6 ␮m to 1 ␮m. Young’s modulus and hardness on the nano scale were assessed with an Ultra Nanoindenter (CSM Instruments SA, Peseux, Switzerland). A load of 8 mN at a speed of 8 mN/min was applied on the sample surface. For each sample, 10–20 indentations were done with a distance of 40 ␮m between adjacent indentations. Poisson’s ratio was set to 0.3 as the literature data for Poisson’s ratio for the ZrO2 –SiO2 system and similar ceramics ranges between 0.25–0.31,8,9 and the difference in Poisson’s ratio would affect the elastic modulus to an extent less than 0.3%. The Oliver–Pharr method10 was used to calculate Young’s modulus, according to: E=

1 − ν2 (2β/Su ) (A/π) − (1 − νi2 )/Ei √

(3)

where ν, Poisson’s ratio of test material; β, Oliver–Pharr constant; Su , slope at start of unloading curve; A, indenter area function (a function of depth of penetration and indenter bluntness); νi , Ei , Poisson’s ratio and Young’s modulus of indenter material. Indentation on the micro scale was done using a micro hardness tester (Buehler Micromet 2104, Lake Bluff, IL, USA) equipped with a Vickers diamond. The DIN EN 843-4 standard11 was used for the test. A load of 2 kg was used for making at least 10 indentations per sample. Indentation diagonals were measured optically using an Olympus AX70 light microscope (Olympus Corp., Tokyo, Japan), at a magnification of 50×. Tip cracks were measured by generating a picture from an Infinity X CCD camera connected to the microscope (Lumenera Corp., Ottawa, ON, Canada), together with the associated software. Cracks were measured within approximately 30 min from when the applied load was removed. Fracture toughness was calculated from Young’s modulus obtained from the nano indentations and the Vickers hardness obtained from the micro indentation, together with average crack lengths as measured optically. The equation proposed by Niihara et al.12 for Palmqvist cracks (l/a < 2.5) was used for the calculations:    √ E 2/5 l −1/2 Kc = 0.018Hv a Hv a 

(4)

where a is the indentation diagonal; E is Young’s modulus, Hv is the Vickers hardness; l is the crack length from the indent tip. The crack profile was characterized by step-by-step polishing away the Vickers indent using a 1 ␮m diamond paste and observing the indentation profile under the microscope. Since the cracks could be polished away while the indentation mark was still visible, i.e. the cracks did not extend under the indentation, the cracks were established to be in the Palmqvist crack system. Measurements were made on samples containing 30, 35 and 40% ZrO2 , as well as a reference specimen of IPS e.max® CAD. Statistical analysis was performed using IBM SPSS Statistics 19.0. Analysis of variance (ANOVA) with Tamhane’s post hoc

Fig. 1. Transmittance over the visible spectra showing a sample containing 35% ZrO2 compared to IPS e.max® . Inserts show (a) appearance of a sample containing 35% ZrO2 and (b) IPS e.max® CAD sample.

test for multiple comparisons was used to evaluate differences at a significance level of 0.05. 2.2.4. Corrosion resistance Corrosion studies were preformed according to ISO 6872.6 Two samples containing 35% ZrO2 were studied. The samples were crushed in order to increase the surface area, washed with distilled water and dried in 150 ◦ C for 24 h. The surface area of the sample was measured using Brunauer–Emmet–Teller (BET) analysis of N2 adsorption isotherms with an ASAP 2020 Physisorption Analyzer (Micromeritics, Norcross, GA, USA). The samples were then weighed and transferred to a glass bottle containing 100 ml acetic acid (4%) preheated to 80 ◦ C. The samples were kept in the acetic acid at 80 ◦ C for 16 h, rinsed with water, dried in 150 ◦ C for 3 days and then weighed again. 3. Results 3.1. Translucency Representative samples of IPS e.max® CAD and the ZrO2 –SiO2 glass ceramic, together with the results of the transmittance study are shown in Fig. 1. The ZrO2 –SiO2 sample was found to be more translucent than IPS e.max® CAD (which is coloured), presenting over 70% transmittance in the visible spectra. 3.2. Crystallinity Bulk XRD analysis was performed on samples containing 30, 35 and 40% ZrO2 , sintered at 1100 ◦ C for 10 or 15 h. Only tetragonal ZrO2 peaks were found, along with SiO2 in some instances, as shown in Fig. 2. The results of the crystallite size calculations for samples containing 30, 35 and 40% ZrO2 are shown in Table 1.

4108

C. Persson et al. / Journal of the European Ceramic Society 32 (2012) 4105–4110

Fig. 2. XRD spectrums of samples containing 30, 35 and 40% ZrO2 , sintered at 1100 ◦ C for 10 or 15 h.

The results of the study on a sample before and after grinding are shown in Fig. 3. No transformation of the tetragonal ZrO2 was noted due to grinding and polishing. 3.3. Mechanical properties Young’s modulus (E), nano- and microhardness (H) and the fracture toughness (Kc ) resulting from the nano- and microindentations are reported in Table 2. The ANOVA found a significant difference between groups for all output parameters (p < 0.001). Tamhane’s multiple comparison test revealed a statistically significant difference (p < 0.01) between IPS e.max® CAD and all other materials for all parameters except Kc for materials containing 30% ZrO2 after 10 h sintering (p = 0.89) and 35% ZrO2 after 15 h sintering (p = 0.99). Young’s modulus was generally lower for the ZrO2 –SiO2 system, whereas the hardness was higher. For the fracture toughness, values both lower and higher than IPS e.max® CAD was found for the ZrO2 –SiO2 system. For the materials synthesized in the current study, Young’s modulus increased with an increase in ZrO2 content, but decreased with an increase in sintering time. These effects were all significant, except 35% ZrO2 (10 h) vs. 40% ZrO2 (10 h), where p = 0.057, and 30% ZrO2 (15 h) vs. 35% ZrO2 (15 h), where p = 0.089. Table 1 The crystallite sizes calculated using Scherrer’s formula. Material 30% ZrO2 –70% SiO2 10 h sintering 15 h sintering 35% ZrO2 –65% SiO2 10 h sintering 15 h sintering 40% ZrO2 –60% SiO2 10 h sintering

Fig. 3. XRD results for a sample containing 35% ZrO2 (specimen from group 3) before and after grinding and polishing. Only tetragonal ZrO2 peaks were found.

For the nano hardness, the sintering time had no statistically significant effect. However, the amount of zirconia did have a significant effect, with higher amounts giving an increased hardness (p < 0.02). Similarly to the nano hardness, no significant effect of sintering time was found for the micro hardness. However, a significant effect due to the amount of zirconia was found (for a sintering time of 10 h only), where 30% ZrO2 gave significantly higher hardness than 35% (p = 0.019) and 40% ZrO2 (p < 0.001). For the fracture toughness, there was a tendency for a decrease in the same with an increase in the amount of zirconia, and an increase with an increase in sintering time. However, in terms of zirconia content, significant differences were found only for 40% ZrO2 , which had significantly lower fracture toughness than all other groups. No statistically significant differences were found due to the sintering time. 3.4. Corrosion resistance The samples were found to lose a very small amount of weight, 0.1 and 3.6 ␮g/cm2 , respectively during the corrosion study. This weight loss could be due not only to corrosion but also to loss of small pieces of material during the transfer of a sample from one beaker to another. 4. Discussion

Crystallite size (nm) 29 33 28 23 35

In this work, translucent, high-strength zirconia–silica glass ceramics with nano-sized grains were produced through a sol–gel process. The XRD results showed only the desired tetragonal peaks, in accordance with previous studies where a sintering temperature of 1100 ◦ C has been used.3,7 When the temperature reached 1150 ◦ C, the monoclinic phase could be observed (data not shown). Previous studies have also reported a transformation between 1100 and 1200 ◦ C for similar compositions.3,13 The

C. Persson et al. / Journal of the European Ceramic Society 32 (2012) 4105–4110

4109

Table 2 Mechanical properties resulting from the nano- and microindentations. Average values and standard deviations are presented. Material

Enano (GPa)

IPS e.max® CAD 30% ZrO2 –70% SiO2 10 h sintering 15 h sintering 35% ZrO2 –65% SiO2 10 h sintering 15 h sintering 40% ZrO2 –60% SiO2 10 h sintering

107.2 ± 3.7

Hnano (GPa)

√ m)

Hmicro (GPa)

Kc (MPa

7.70 ± 0.51

6.07 ± 0.07

3.09 ± 0.11

88.7 ± 1.2 83.8 ± 0.7

9.51 ± 0.39 9.68 ± 0.28

8.01 ± 0.11 7.93 ± 0.28

3.49 ± 0.96 4.27 ± 0.93

96.3 ± 3.5 86.9 ± 2.3

10.01 ± 0.76 12.59 ± 0.61

7.74 ± 0.19 8.03 ± 0.31

2.63 ± 0.54 3.30 ± 0.73

99.7 ± 7.8

10.85 ± 1.60

7.69 ± 0.26

2.30 ± 0.30

particles sizes were also below the critical size of 40 nm found previously.3 Contrary to the study by Nogami and Tomozawa,3 no tetragonal–monoclinic transformation was noted due to grinding. This would be advantageous for post-treatment of this type of material, such as re-shaping, grinding and polishing. Nogami et al. reported that the high fracture toughness of this material was due to the tetragonal–monoclinic transformation. However, Wang et al. reported that the strengthening and toughening mechanism may mainly be due to residual stresses in the ceramic grain boundary rather than the tetragonal–monoclinic transformation in a ZrO2 –SiO2 system with nano-grain size.14 The nano-sized ZrO2 tetragonal phase could be stabilized by the silica matrix due to its constraint and the formation of a Zr–O–Si interlayer surrounding the ZrO2 grains. Wang et al. also reported that the monoclinic phase could be observed at the interlayer between ZrO2 and SiO2 under a certain stress, but it could not be observed by XRD. For the mechanical properties, the materials studied here presented a somewhat higher hardness but a slightly lower elastic modulus than IPS e.max® CAD. A higher hardness was found with the nanoindentations compared to the microindentations, which may reflect the smaller volume measured and/or a higher hardness of the surface compared to the bulk. The fracture toughness of the materials synthesized in this study was found to lie both below and above that of IPS e.max® CAD, depending on the zirconia content and sintering process. A zirconia content of 30% together with a sintering time of 15 h was found √ to give the highest fracture toughness, 4.27 ± 0.93 MPa m, √ compared to the 3.09 ± 0.11 MPa m of the IPS e.max® CAD. This is somewhat lower than that of Nogami and Tomozawa,3 who produced samples containing twice the amount of zirconia. However, attempts to reproduce such material were unsuccessful, as precipitation occurred instead of gelification. It should be noted that the method of obtaining fracture toughness values (indentation) is not the one recommended by the dental standards (ISO 68726 e.g. recommends the single edge V-notch beam). However, indentation has been frequently used to evaluate and compare dental materials,1,15 and may be adequate as a preliminary indication of the crack growth resistance of the material. Different equations have been suggested for the calculation of the fracture toughness based on indentations.15 In this study, the type of crack created was confirmed experimentally (Section 2.2.3) and found to be of the Palmqvist type. The equation used was chosen accordingly.

The chemical solubility of the synthesized materials was 3.6 ␮g/cm2 or less, which is well below the maximum allowance of 100 ␮g/cm2 cited in the standard for dental materials,6 as well as the 78 ␮g/cm2 reported for lithium disilicate.16 In this preliminary study, promising results were obtained although it should be noted that the process is sensitive to outer disturbance and crack-free samples were difficult to obtain. Further studies will be performed, focusing on optimizing the manufacturing process in order to avoid cracks. Furthermore, alternative process routes will be explored, such as drying the xerogels and sintering the resulting powder. This would be a more costly process, but may give shorter processing times as well as a decreased crack propensity. Future studies should include the possibilities of obtaining different dental shades. Increasing the time and temperature of the heat treatment would provide a white shade but may also alter the mechanical properties. The incorporation of different polyvalent ions into the material, as in lithium disilicates, should also be evaluated for colouring purposes.17,18 5. Conclusions Translucent, high strength glass ceramics in the zirconia–silica system with nano-sized grains were produced using a sol–gel method. The phase composition appeared unaltered after grinding and polishing, and a high fracture toughness in comparison to lithium disilicate could be achieved. The translucency and corrosion resistance were adequate for the requirements of dental materials. Further development of this material for use in dental applications may be hindered by difficulties in achieving larger dimensions due to the crack formation propensity of the material. References 1. Yilmaz H, Aydin C, Gul BE. Flexural strength and fracture toughness of dental core ceramics. J Prosthet Dent 2007;98:120–8. 2. Luthy H, Filser F, Loeffel O, Schumacher M, Gauckler LJ, Hammerle CH. Strength and reliability of four-unit all-ceramic posterior bridges. Dent Mater 2005;21:930–7. 3. Nogami M, Tomozawa M. ZrO2 -transformation-toughened glass-ceramics prepared by the sol–gel process from metal alkoxides. J Am Ceram Soc 1986;69:99–102. 4. Lohbauer U, Muller FA, Petschelt A. Influence of surface roughness on mechanical strength of resin composite versus glass ceramic materials. Dent Mater 2008;24:250–6.

4110

C. Persson et al. / Journal of the European Ceramic Society 32 (2012) 4105–4110

5. Nogami M. Glass preparation of the ZrO2 SiO2 system by the sol–gel process from metal alkoxides. J Non-Cryst Solids 1985;69:415–23. 6. ISO 6872. Dentistry – ceramic materials; 2008. 7. Miranda Salvado IM, Serna CJ, Fernandez Navarro JM. ZrO2 –SiO2 materials prepared by sol–gel. J Non-Cryst Solids 1988;100:330–8. 8. Makishima A, Mackenzie JD. Calculation of bulk modulus, shear modulus and Poisson’s ratio of glass. J Non-Cryst Solids 1975;17:147–57. 9. Guazzato M, Proos K, Quach L, Swain MV. Strength, reliability and mode of fracture of bilayered porcelain/zirconia (Y-TZP) dental ceramics. Biomaterials 2004;25:5045–52. 10. Oliver WC, Pharr GM. An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J Mater Res 1992;7:1564–83. 11. DIN EN 843-4. Mechanical properties of monolithic ceramics at room temperature; 2005. 12. Niihara K, Morena R, Hasselman DPH. Evaluation of KIc of brittle solids by the indentation method with low crack-to-indent ratios. J Mater Sci Lett 1982;1:13–6.

13. Aguilar DH, Torres-Gonzalez LC, Torres-Martinez LM, Lopez T, Quintana P. A study of the crystallization of ZrO2 in the sol–gel system: ZrO2 –SiO2 . J Solid State Chem 2001;158:349–57. 14. Wang S-W, Huang X-X, Guo J-K. Mechanical properties and microstructure of ZrO2 –SiO2 composite. J Mater Sci 1997;32:197–201. 15. Ponton CB, Rawlings RD. Vickers indentation fracture toughness test. Part 1. Review of literature and formulation of standardised indentation toughness equations. Mater Sci Technol 1989;5:865–72. 16. Fathi H, Johnson A, van Noort R, Ward JM, Brook IM. The effect of calcium fluoride (CaF(2)) on the chemical solubility of an apatite-mullite glass-ceramic material. Dent Mater 2005;21:551–6. 17. Anusavice KJ, Zhang NZ, Moorhead JE. Influence of P2 O5 , AgNO3 , and FeCl3 on color and translucency of lithia-based glass-ceramics. Dent Mater 1994;10:230–5. 18. Anusavice KJ, Zhang NZ, Moorhead JE. Influence of colorants on crystallization and mechanical properties of lithia-based glass-ceramics. Dent Mater 1994;10:141–6.