Advanced Ceramics - Norton Company Northboro, MA 01532

Transcription

1 DETECTION OF CRACKS IN Si 3 N 4 BASE CERAMICS USING MICROFOCUS RADIOGRAPHY Kamal E. Amin Advanced Ceramics - Norton Company Northboro, MA 1532 BACKGROUND The long term reliability of structural ceramic components is of major interest and concern to those involved in design, manufacturing, testing and use of such components [1,2]. Flaw population data generated at a high probability of detection and high confidence level using improved NDE technology undoubtedly provide a reliable method for safe design at a lower cost as illustrated by the concept in Fig. 1. The concept demonstrates that more flaws with smaller size distribution may be detected at higher probability/ confidence level (compare dotted areas under the two flaws distributions 1, 2) with the use of appropriate NDE techniques and technique development. Defects usually encountered during processing of ceramics include cracks, inclusions, voids, density gradients and residual stresses [3]. Cracks are usually of major concern since they significantly reduce the fracture strength of such materials. Fracture usually starts at a preexisting crack in the material under the influence of concentrated stress at the crack tip [4]. For volume cracks: K 1 c o..jlfa f (a/b) For surface cracks: K 1 c a..j1fa Such stress concentration, K 1 c, is usually modified by the presence of residual grinding and thermal stresses. The two most commonly used NDE techniques for inspection and flaw detection in industry today are radiography and ultrasonics [5). Review of Progress in Quantitative Nondestructive Evaluation, Vol. 9 Edited by D.O. Thompson and D.E. Chimenti Plenum Press, New York,

2 [/) +- Q) 4- Q) 4- c. +- Q) + Q) 4- ::::Jl.D (!).D L [}.9 Probabili tl:) of Detection Flaw Distribution ( N VS L l Flaw Size ( NDE l Fig. 1. Design flaw size based on NDE (at 9% POD) 1 - Processing Iteration #1 technology A - NDE existing 2 - Improved Processing Iteration #2 B - NDE technology improvement Literature data reveal very limited work on crack detection using radiography techniques. J.J. Mecholsky et al [6] during a study on crack propagation in selected ceramic materials (glasses, MgF 2, graphite, Al 2 3, hot pressed Si 3 N 4 ) showed that individual cracks with spacing <7 ~ m were not detected by microfocus film radiography technique. The cracks they generated were at least few millimeters long, with widths of around 3 ~ m of crack measurements. or larger. Their work did not specify the method N. Lapinski and A. Sather have a patent [7] describing a method to enhance crack detection through immersing the part in a solution of AgN 3 which contains appropriate surfactant. They claim detection of a crack around 1 ~ m Si 3 N 4 deep in No mention was made for how the cracks were measured and what procedure were used to ensure repeatability of both inducing and detecting cracks with known dimensions. Much literature has been published on detection of cracks using ultrasonic techniques [5]. Isolated cracks are easily detected using either surface waves, shear waves or longitudinal waves. However, data processing and analysis demands the availability of complex software for signals obtained from a network of cracks. Fluorescent dye penetrant, acoustic emission and thermal wave imaging may also provide information on cracks in ceramics, they however have serious limitations. The present investigation was conducted for the purpose of quantitatively assessing the detection capability of microfocus film radiography for cracks induced under controlled conditions and fully characterized. This information is critical to setting up inspection procedure of ceramic parts at Norton Company. 1162

3 EXPERIMENTAL DETAILS The material used in the present investigation is a Si 3 N 4 base ceramic containing 4 wt% Y 2 3 densified using hot isostatic pressing. Small tiles 1.6 mm thick were machined from the above material and polished to a.3 ~ m final surface finish. Cracks were then introduced by indenting the samples using a Vickers diamond indentor under normal loads in the range 45 to 785 N; see the example of Fig. 2. In addition, plate samples machined to 25.4 by 76.2 mm and thickness of 1 mm having a surface finish around 15 ~ m were used. They were then notched using a diamond saw and loaded in double torsion using 3 to 36 N force to generate very sharp small cracks, see Fig. 3. Crack dimensions are documented in Table I. Larger cracks were generated by applying indentations (at loads <4 N) at locations near the precracks in those samples as shown in Figs. 4 and 5. Cracks length as large as 12.5 mm with widths up to.6 mm and depths to 1 mm were produced. Crack dimensions were measured from the fracture surfaces and sample dimensions using both optical and scanning electron microscopy, as shown in Fig. 6. Measurements were also confirmed using dye penetrant technique. MICROFOCUS RADIOGRAPHY The HOM X-161 IRT/Ridge microfocus radiography system was used to detect those cracks. It was run at 4 to 9 KV,.32 to.82 ma for exposure times of 3 to 4 minutes. Film radiographs were shot at three different orientations each at a 9 angle to the beam. Both Agfa D2 and D4 films were used and film density of 1.8 to 2.2 was achieved. Samples were examined at magnifications in the range 3 to 18 X. An Agfa structurex film processor was used with cycle time of 8 min. Fig. 2. An example of Vickers indentations array generated in Si 3 N 4 ceramics under 785 N normal load for the radiography study of crack detection. 1163

4 Figure 3. Representative Optical Micrographs of Precracked Si 3N 4 plates. Cracks are:.9 to 1.7 mm long, <.1 in inches and around 3 ~ m deep. Such cracks were not detected by microfocus radiography. Figure 4. Scanning electron micrographs of examples of cracks propagated in Si 4 N 4 plates by introducing indentations close to the precracks locations (see arrows). Those indentations were introduced under loads in the range of 29 to 34N. 1164

5 Figure 5. Examples of the Fracture Surface of Notched and Precracked si 3 N 4 Bars Showing the Location and Depth of Cracks from Both Indentations and Double Torsion Loading. Cracks: Depth around 3 urn: Length 18 urn. and width <2 um. Dummy samples of the same material were added to the plates to achieve thickness range of 3 to 44 mm to determine the thickness sensitivity of crack detection. An example of the effect of sample thickness on crack detection is illustrated in the radiographs of Figs. 6 and 7. RESULTS AND DISCUSSION Radiographs of the samples with Vickers indentation cracks did not reveal any crack contrast although those cracks were as long as.5mm. Those cracks have widths as high as.5 rnm and depths in the range.3 to.5 mm; see Table I. The crack dimension in the direction of the x-ray beam ranges from around.25 % to 31.2 % of the sample size (percent thickness sensitivity). All those cracks are half penny in shape and extend to the surface (Fig. 2). Failure to detect these cracks was a common observation for all the cracks, which were induced in a geometric array in such samples, at various sample orientations to the beam. The cracks induced at the notch of the plate samples (by double torsion loading) did not show any indication on both the slow and medium speed film radiographs (Fig. 6) and with and without the use of appropriate luminescent screen filters, usually introduced to enhance film density and reduce scattered radiations. Such cracks had lengths as high as 1.4mm, widths to.5mm and depths of.3 mm. This translates into percent thickness sensitivities of 3 to 4. The indentations applied at the tip of those cracks provided crack extension for further study. The cracks reached a length of around 1.7 mm to 2.8 mm. Upon increasing the indentation load the cracks propagated immediately throughout 1165

6 the full specimen (at normal loads >294 N). Representative radiographs are shown in Fig. 7 illustrating the effect of sample thickness on crack detection in those plates. The present investigation data reveals the following: Cracks with lengths around 27 to 28.5 % of the specimen thickness and depth of 9% are detectable when their width is around.13% of the specimen thickness. However, when their width is. 45% of the specimen size, they are not detected by microfocus; compare rows 6 and 8 of Table I. On the other hand cracks with width sensitivity of 5% of the specimen size are not detected no matter how long they are if their depth is ~ 3 of % the specimen size.; compare rows 7 and 9 of Table I. Moreover, cracks with width sensitivity of << 1% of the specimen size cannot be detected even at length sensitivity of 31.2% and depth sensitivity of 2%; see row 12 of Table I. Additional examples may be deduced from the table data. Considering such information and using fracture toughness value of 5 M P for a ~ Si 3 N 4 a predicted fracture strength may be computed based on radiography data as shown in Table II for both surface and volume crack. The calculations yield fracture strength data below acceptable values for design, <515 MPa. In other words, quality screening for design based on radiography data of crack detection is not sufficient and other screening NDE techniques are a must if we are to reduce or eliminate the expensive proof testing procedure. Looking at the table we may observe the fact that radiography provides better detection data for inclusions and voids/pores from which the predicted fracture strength is useful for specific applications. In many applications a guaranteed fracture strength of over 55 MPa is essential. Notched precracked Figure 6. Examples of Microfocus Radiographs of Notched Plates Before and After Cracking in Double Torsion Under Around 31 N force. The cracks are <2 urn Wide and were not Detected by Microfocus Radiography. 1166

7 Figure 7. Microfocus Film Radiographs Showing Crack Detection as a Function of Plate Thickness in HIP'ed Si 3 N 4 ceramics. Plate thickness: mm; b - 2.3; c mm Table I Ct... DIMeMIOM... Creek LIT W/T D/T Creotl DlmbltonX 1 In Ttllc*n" Det1ctlon X X X /Ttah3n.. Retlo In w D Direction 1.., tu u y.ii 21., 21., u y u ".u 21., 21., y u.s. 31 Jt.l Jt.l 12.1 " y "... Jt.l Jt.l y , " y 21.5.u,, 12.5 " " 31.1 y N 27..., 1.7 ".u N u N... ' "....u N " o.ao.4 o.a 1.8 N , 2....tu... 1., N oo.3... N It.It II.... " 12 t.l.tu , N '-3.u,,3.u.,.3, 4... ' ''.tu.ttj ".14 1., N 2. 5.u 2. 5.u. 2.5, 1., N 5.u.lt. 5, , N 3.1.u 3.1.It, J.l, Sufac. Cracl "' ntd.ani S.ulthlt:p+ '' ( 1 ) ( 31 ) - Table II...,.,. Pnh.Wb' I J5 c.arw... ::. Ztt a-... PrMJ- r.. s.auoot l'lo... Dot - ( Ut).,7 (,,_. ) II 3 ( 16.4 ) v... Crack 6'' Tllcb.,. S u:ltlfltj+ ft ( I.. tl ) lttt<h-,1 ) > ltt 514 (..., ) 116 ( ll ) " 47., > 2tt 117 ",,. ( u 3 ) 512 ( 14.5 ) 7JJ(IIf) + Oreok width 8 urn. eharper oreoke ( width < 2 um ) thiokn... eeneltlvlty ie >1". 1167

8 From our experience we find that microfocus radiography is useful as a screening QC tool during processing to reveal gross defects in green ceramics as well as at the latter stages of processing. However, other NDE techniques such as ultrasonics are useful for detecting flaws of interest to designers. CONCLUSIONS 1. Total crack geometry should be considered when using radiography detection techniques. 2. The minimum crack dimension in the direction of the x ray beam must be >5% of the part thickness for detecting both surface and volume cracks. Other crack dimensions must be >.5% of the part thickness. 3. Ultrasonics and other NDE techniques are recommended for detecting those cracks in the critical flaw size range which cannot be detected using x-ray radiography. REFERENCES 1. A.G.Evans, "Structural Reliability: A Processing-Dependent Phenomenon", J. Amer. Cer. Soc., Vol. 65, No. 3, 1982' p D.R.Johnson, R.W.McClung, M.A.Janney and W.M.Hanusiak, "Needs Assessment for Nondestructive Testing and Materials Characterization for Improved Reliability in Structural Ceramics for Heat Engines", Oak Ridge Nat'l Lab. Tech. Rpt. #ORNL/TM K.E.Amin, T.Derkacs and S.Matay, "Detectability of Seeded Defects in HIP'ed Si 3 N 4 Using Both Ultrasonics and Radiography, Proceedings of the 3rd Int'l. Symp. on Ceramic Materials & Components for Engines, Las Vegas, September, s.w.freiman, "Brittle Fracture Behavior of Ceramics", Ceramic Bulletin, Vol. 67, No. 2, 1988, p L.R.Clarke, B.T.Khuri Yakub and D.B.Marshall, "Resolution of Closely-Spaced Machining-Damage-Induced Surface Cracks in Ceramics", p. 1617, Proceedings of Review of Progress in Quantitative Nondestructive Evaluation, Vol. 6B, Ed. DOThompson and DEChimenti, Plenum Press, NY and London. 6. C.C.M.Wu, S.W.Freiman, R.W.Rice and J.J.Mecholsky, "Microstructural Aspects of Crack Propagation in Ceramics", J. Mater. Science 13, 1978, p N.Lapinski and A.Sather, u.s. Patent# 4,172,224, "Process for the Detection of Micro-cracks", oct. 23,