Crack propagation and residual stress in laminated Si3N4/BN composite structures

https://doi.org/10.1016/j.jeurceramsoc.2011.03.009Get rights and content

Abstract

The level of residual stress and crack propagation in a new generation of laminates, based on silicon nitride (Si3N4) layer and a mixture of boron nitride (BN) and alumina (Al2O3) interlayer, was presented. The structure consists of alternated concentric rings of Si3N4 separated by the weak BN interlayer possessing no planes of easy crack propagation and fracture resistance much larger than that of any classical planar laminates. The results on direction of crack propagation and residual stress in relation to inter-layer composition, the number of layers, and their thickness are investigated and reported. The effect of residual stress on crack propagation was studied by using Vicksrs intentation. The highest compressive residual stress of ∼170 MPa was found in samples with five layers possessing an average layer thickness of ∼310 × 10−6 m.

Introduction

A new generation of laminated structures has been designed and fabricated using a modified slip casting method followed by pressureless sintering.1, 2, 3 Instead of layers being deposited in the planar form, as is the case with planar laminates, in the new design they are arranged in the form of concentric rings along the length of the structure thus avoiding the creation of plane for easy crack propagation.2

Two different laminate compositions (in the terms of interlayer phase composition) were designed. The first one, consists of Si3N4 layers, and interlayers containing 10 wt.% of Si3N4 and 90 wt.% of BN marked as SN  (SN + BN); and the second one, consists of Si3N4 layers, and interlayers containing 50 wt.% of Al2O3 and 50 wt.% of BN marked as SN  (BN + Al2O3). In these types of structures the objective is to create interfacial layer sufficiently weak to allow crack deflection along the interface. On the other hand, the strength of the Si3N4 layer must be sufficiently high to prevent an easy crack initiation at the surface of the next layer. The residual stress is known to strongly affect mechanical behaviour of the laminated composite by controlling the crack propagation at the interface between the two dissimilar materials.4, 5 According to He et al., there are three possibilities for the crack propagation: the crack may arrest at the interface under compressive stress, it may penetrate the interface, or it may deflect from the interface. The structural model which will provide this to take place is proposed in Fig. 1.

During the fabrication of composites with two dissimilar materials (Si3N4 layers and BN  Al2O3 interlayer) residual stresses will arise due to the mismatch in thermal expansion and differences in elastic properties.6, 7, 8 In the case of laminates with different ceramics, the differences in shrinkage during sintering will also contribute to the residual stress.9 Perhaps, the most important factor which contributes to the rise in residual stress is the difference in the linear coefficient of thermal expansion between the two materials used to fabricate laminates. The residual strain, responsible for the creation of residual stress, which develops in a composite laminate consisting of materials with different thermal expansion coefficients, as proposed by the Oechsner et al.10 can be expressed by equation:εM=TT0(α2α1)dTwhere α1 and α2 are the thermal expansion coefficients of the two materials and dT is the temperature difference T1  T0.

In the Si3N4 laminates the residual bi-axial compressive stress in the Si3N4 is given by Blugan et al.11:σC=εME11+(t1E1/t2E2)where E = E/(1  ν), E1 and E2 are Young's modulus of materials with thermal expansion coefficients of α1 and α2, respectively, ν is the Poisson's ratio and t1 and t2 are the thickness of the layer 1 and layer 2, respectively. In the interlayer (Al2O3 + BN) which has larger thermal expansion coefficient compared to silicon nitride, the bi-axial tensile stress is given by11:σT=σCt1t2If t1/t2 approaches 0 then the bi-axial stress in material with lower thermal expansion coefficient (i.e. Si3N4) reduces to:σC=εME11νand the stress in the material with larger thermal expansion coefficient vanishes (σT approaches 0).

The effect of residual stress on crack propagation was studied using Vickers indentation by Hsueh and Evans6 who developed a method of measuring the residual stress in a ceramic matrix by comparing the growth of crack from indentations made in unstressed and stressed materials. The same method was used by Marshal and Lawn12 to evaluate the residual stress in tempered glass. In 1994 Zeng and Rowcliffe13 developed a method of mapping the residual stress in glass. First, they introduced a residual stress in the glass by applying high external loads. Then, using low loads, small indentations were made at various angles and distances from the cracks extending from the large indentations. It was than possible to map the residual stress field created on the materials surface by the indentation.

By using indentation fracture mechanics theory it has been possible to determine the fracture toughness, KIC, of the material by measuring the load and crack length C0, via equation of the form14:KIC=αEH1/2PC03/2where P is the indentation loads, α is an empirical geometric constant, and E and H are Young's modulus and hardness, respectively. Following Zeng and Rowcliffe13 it has been shown that, under the influence of a prevailing residual stress, σa, the crack will assume a new equilibrium length, C, when the surface of the specimen is loaded at the same indentation load, P, as in the unstressed case. At equilibrium, the crack will then experience composite stress intensity, described by the fracture toughness, KIC, and is given by the expression:KIC=αEH1/2PC03/2±ψσaC1/2The first term in Eq. (6) represents the stress intensity due to the indentation load, P, and the second term corresponds to the contribution of a prevailing residual stress. The sign ± refers to the tensile or compressive stress. Under the tensile stress, the second term is added to the first term, and under the compressive stress the second term is subtracted from the first term. In Eq. (6), ψ is a crack geometry factor and describes the nature of the surface-to-depth ratio of the crack dimensions. The commonly used value for ψ for Vickers indenter is π (Lawn).15 Combining Eqs. (5), (6) and noting that the same peak load is involved in both expressions, the unknown residual stress is obtained through the expressions:σT=KIC1(C0/C)3/2ψCσC=KIC1(C0/C)3/2ψCwhere σT and σC represent residual tensile and compressive stresses, respectively. A tensile residual stress will act to extend the stress-free crack so that C > C0, while a compressive residual stress will shorten it so C < C0.

The objective of this paper is to investigate the effect of number of layers, their thickness and thermal properties on the level of residual stress developed in the layers of the laminates. Also, an additional objective of this paper was to study the role of the residual stress in toughening of the laminated composites.

Section snippets

Experimental procedure

Si3N4/BN based laminates were slip-cast alternately with Si3N4 layers and BN based interfaces in plaster of Paris mould in a casting chamber of rectangular/square cross-section of 8 mm × 8 mm and ∼60 mm deep. All laminates were fabricated in such a way that the outside layer and core are Si3N4. Two types of laminated materials were fabricated and studied in this paper. One consists of Si3N4 layers and BN based interface with 90 wt.% of BN and 10 wt.% Si3N4 marked as SN  (SN + BN). The other one consists

Results and discussion

In the Si3N4/BN laminates the material with lower thermal expansion coefficient is the Si3N4 with α = 3.2 × 10−6/°C. The phase with higher thermal expansion is the interfacial YAG phase2 which is formed as a result of reaction between Al2O3 and Y2O3 and has the thermal expansion coefficient of ∼8 × 10−6/°C. Substituting appropriate values for νSN = 0.22, ESN = 320 GPa, αSN = 3.2 × 10−6/°C, the average layer thickness of t1 = 230 × 10−6 m, νiterf. = 0.25, Einterf. = 390 GPa, αinterf. = 8 × 10−6/°C, t2 = 15 × 10−6 m and ΔT = (1300  

Conclusion

The level of residual stress developed in the interface was found to depend on the nature/composition of the interface and Si3N4 layer thickness. The compressive stress of 170 MPa, was found to exist in higher density inter-layers of SN  (BN + Al2O3) laminates compared to the stress of 75 MPa in lower density inter-layers of SN  (BN + SN) laminates. Both, the residual stress and the interfacial strength govern the direction of crack propagation which, in turn, controls the fracture resistance of the Si3

References (17)

There are more references available in the full text version of this article.

Cited by (14)

  • Fabrication of ceramic sealing coatings for shell bionic structures and their failure mechanism during thermal cycling

    2023, Ceramics International
    Citation Excerpt :

    The high strength and toughness of shell nacre not only benefit from the alternate connection of the hard phase and soft phase but also depend on the microstructure formed by the interaction and orderly arrangement between the two phases. Krstic et al. [29] designed and manufactured a new type of Si3N4-based layered structure composite by the gel injection moulding method and studied the effect of layer numbers on fracture toughness. When the numbers of Si3N4 layers were 7 and 4, the fracture toughness of the Si3N4-(BN + Al2O3) composite and Si3N4-(BN + Si3N4) composite reached the best values, respectively.

  • Effect of pores on crack propagation behavior for porous Si<inf>3</inf>N<inf>4</inf> ceramics

    2016, Ceramics International
    Citation Excerpt :

    Few engineering materials are limited by their strength; instead, they are limited by their resistance to fracture or fracture toughness. In particular, the fracture issue due to internal crack propagation has drawn the attention of many researchers [5–7]. Experimental results indicate that the crack propagation is controlled mainly by the crack tip stress intensity factors K [8–11].

View all citing articles on Scopus
View full text