Search

Disc and toroidal shaped samples of composites made from powders of Na-β-alumina and tin-stabilised tetragonal zirconia by conventional mixing and sintering processes have been used to establish data on the effect of the zirconia on the sintering efficiency, the mechanical strength, hardness and fracture toughness. These data have been compared to the results obtained when the β-alumina powder was first pre-sintered and then infiltrated with solutions of zirconium salts or suspensions of the tin-stabilised tetragonal zirconia. The results indicate that significant improvements to the mechanical properties can be made when the zirconia is introduced via an organometallic salt dissolved in an organic solvent. For the conventionally made samples the strength and toughness peak at 15 wt% zirconia addition. Sample densities in excess of 97% are achieved by pressurless sintering at temperatures up to 1600 °C.

The stresses for macroscopic plastic flow and critical stages of fracture, fracture toughness and hardness of sintered plus forged T1 high speed steel were determined. The results are compared to similar data for sintered, sintered to closed porosity plus hot isostatically pressed and electroflux refined (EFR) alloys of comparable composition. EFR meltstock, with addition of 0.6 wt% Mo, was water-atomized in a 200 kg unit which incorporated ceramic filters and an argon shroud to ensure maximum cleanliness. The powder was sieved

R-curve behaviour in the microstructurally short crack regime has been reported mainly in ceramics, composites and polymers; this paper describes it for a metallic material: brittle cast and wrought T42 high speed steel. Continuum mechanics has demonstrated the general admissibility of sub-critical growth of cracks with a cohesive zone. Results now reported show that, in a metallic material, growth of microstructurally short cracks under monotonic loading, as in fatigue, is governed by microstructure (texture). Natural cracks, i.e. produced by the hot forging operation, or induced by the application of stress in the range 0.5 to 1.1 GPa in four-point bending experiments, of depths extending to 25µm were always associated with MC carbides. At comparable stress levels cracks were nucleated in compression -- surprisingly some transverse to the compressive axis. Observations of crack nucleation and subsequent studies of subcritical growth of these microcracks were made by surface replica microscopy. Crack extension was easy within the carbide stringers (a characteristic feature of hot-worked high speed steels), but, at higher stresses, took place between these bands to reach up to ∼ 100 µm (surface) length. Dormant cracks were shallow, no more than 6 µm deep; whereas those responsible for failure, at stresses ranging from 0.6 to 1.9 GPa, had a semicircular geomerty -- identified by scanning electron fractography. Step-wise monotonic subcritical crack growth is modelled asR -curves and it is shown that the maximum estimated (microscopic) applied stress intensity factor K a can vary from 0.5 to 1.0 K1C, the macroscopic fracture toughness independently determined using sharp artificial long cracks.

The tensile and compressive strengths of three polyester resins were measured under superposed hydrostatic pressure extending to 300 MPa, in an attempt to establish yield criteria. The polyesters were brittle in uniaxial tension at all pressures, and accordingly, a third testing geometry, diametral compression of a disc, was employed to complete the two or three necessary parameters to establish the yield surface in stress space. From the biaxial (disc) and axial compressive test data, the atmospheric tensile yield strength (higher than the fracture strength) was computed to be ∼67 MPa in comparison with the compressive strength of ∼120 MPa, their ratio 0.56 being significantly less than the more common 0.75 found for thermoplastics and epoxides. The data for compressive yield strength under superposed pressure were compared with the predictions of the two-parameter pyramidal, conical and paraboloidal criteria and the fit, though reasonable for the latter, could be significantly improved if a further independent material parameter was employed to give a three-parameter pyramidal criterion (the principal stresses σ1, σ2 and σ3 being measured in MPa) of the form 0.0150σ1−0.0039σ2−0.0083σ3=1

An attempt was made to measure indirectly the transverse tensile strengths of uniaxially aligned fibre pultrusions by the diametral compression of disc-shaped samples using concave loading anvils. Two types of composite were investigated, containing ∼60%Vf of either type AS carbon fibres (CFRP) or S glass fibres (GRP), both in an epoxy resin matrix. Testing was carried out at atmospheric and under superposed hydrostatic pressures, −H, extending to 300 M Pa. The resultant principal stresses at the disc centre were δ1 = δA +H; δ2 =H; δ3 = −3δA +H, where δA = 2P/πdt for a disc of diameter,d, and thickness,t, subjected to a loadP. Deviations from linearity in the load-deflection response were detected throughout the pressure range at a70% and a90% of the failure load for CFRP and GRP, respectively, and these were associated with resin yielding. The pressure dependence of δA, approximately −0.1H, was consistent with a two-parameter yielding criterion predicting hypothetical yield stresses in simple tension and compression of a81 and −109 MPa, respectively, for both matrix materials. Irrespective of pressureeventual fractures took place along the loading diameter, but in the CFRP specimens tested under pressure initial cracks at the disc centres were at a45 ° to the loading axis, i.e. on the plane of maximum shear stress. Fractographic observations were consistent with transverse failure taking place by fibre-matrix decohesion in GRP and by resin fracture in CFRP. Other than the atmospheric datum point for CFRP, the pressure dependence of δA for failure, δF, was also approximately −0.1H. Of the various stress, strain and strain energy criteria for failure examined, only critical tensile strain was found consistent with this pressure dependence.

Single crystals of LiF containing voids and of NaCl containing Na2SO4 precipitates were pressurized to introduce dislocations in the vicinities of the discontinuities and subsequently compressed along 〈100〉 at room temperature. The yield stress was raised in both materials; additionally, in LiF discontinuous yielding and easy glide were suppressed and work hardening rate increased by the pressurization-induced dislocations. Following pressurization at 0.85 GN m−2, for example, the 0.1 % shear flow stress of LiF was doubled to ∼ 4 MN m−2 and stage II work hardening rate quadrupled to ∼ 180 MN m−2. Pressurization of NaCl above 0.6 GN m−2 resulted in an increase in the 0.1 % flow stress from ∼ 1.2 to ∼ 2.0 MN m−2. If the slip bands in LiF were initiated by a precompression, pressurization prevented the broadening of these fresh slip bands during subsequent plastic flow. Deformation now took place at a higher stress both in LiF and NaCl. These effects resemble in some ways latent hardening in that oblique as well as conjugate dislocation intersections must take place to continue the deformation. In contrast to latent hardening data, the strain hardening rate was increased in LiF and was approximately proportional to the pressurization-induced dislocation density. This ratio, 5 to 6 dyne per dislocation, is in fair agreement with two sets of independent calculations reported by Gilman and Johnston. The results suggest, therefore, that in the present case also hardening may be due to defects left in the wakes of pressurization-induced moving dislocations.

Initiation and propagation of failure in uniaxially aligned 60% volume-fraction Type III carbon fibre-epoxide compressive specimens, strained parallel to the fibre axis, was studied at atmospheric and superposed hydrostatic pressures, H, extending to 300 MN m−2. The atmospheric axial compressive strength was approximately 1.5 GN m−2 and equal to the tensile strength, but mechanisms involving shear-operated failure of the fibres must be discounted since the failure process was very pressure sensitive above H∼ 150 MN m−2. The results also could not be satisfactorily interpreted by theories involving micro-buckling of individual fibres or laminae when the matrix shear modulus controls the compressive strength. For atmospheric tests and for H

The failure mechanisms in waisted tensile specimens of pultruded 60% volume fraction glass fibre-epoxide were investigated at atmospheric and superposed hydrostatic pressures extending to 350 MN m−2. The maximum principal stress at fracture decreased from ∼1.7 GN m−2 at atmospheric pressure to ∼1.3 GN m−2 at 250 MN m−2 superposed pressure and remained approximately constant at higher pressures, as had been observed with carbon fibre reinforced plastic (CFRP) and a nickel-matrix carbon fibre composite. In the high-pressure region the failure surfaces were fairly flat, consistent with the fracture process being solely controlled by fibre strength. Pre-failure damage, in particular debonding, was initiated at ∼0.95 GN m−2 at atmospheric pressure and this stress rose to ∼1.2 GN m−2 at 300 MN m−2 superposed pressure, i.e. by about 9% per 100 MN m−2. Unlike the pressure dependence in CFRP, this contrasts with the pressure dependence of the resin tensile strength, about 25% per 100 MN m−2, but can be associated with that of the fibre bundle/resin debonding stress, about 12% per 100 MN m−2 superposed pressure. Consistent with this interpretation, glass fibres of the failure surfaces were resin-free, again in contrast to CFRP.

The tensile, compressive and shear yield strengths of two epoxides were measured under superposed hydrostatic pressure extending to 300 MN m−2. For both materials, the ratio of the moduli of the tensile,σT, to compressive,σC, yield stress at atmospheric pressure was approximately 3∶4, as has been reported previously for a number of thermoplastics. Theσ2=σ3 envelope in stress space was plotted according to these two-parameter (σC andσT) yield criteria: conical, paraboloidal and pyramidal; the best correlation was with the last. The experimental tensile and compressive data for tests under pressure, however, fit slightly better two straight lines which are consistent with a three-parameter single hexagonal pyramidal yield surface. For plane stress and shear under pressure yield envelopes of these surfaces, the correlation with experimental data is again best for the pyramidal criteria, except for biaxial or triaxial tension when these resins are brittle. The third independent parameter employed in the pyramidal criterion was the equi-biaxial compressive yield stress, determined by tensile experiments under appropriate superposed hydrostatic pressure; alternatively plane strain compressive yield stress,σPC, may be used.

The first stage of the failure process in pultruded, 60% volume fraction, type III carbon fibre-epoxide beam specimens with span-to-depth ratios of 5, 15 and 40 deformed in flexure at atmospheric pressure was the initiation of kinking by the “compression” roller. Kink growth during the non-linear part of the load-deflection curve was followed by kink propagation at peak load. Acoustic emission and load-unload tests to detect irrecoverable deflection supported direct microscopic observations of damage. Kink growth with decreasing load, increasing deflection and accompanying redistribution of stresses led to two types of failure, commonly referred to as “flexural” and “interlaminar”. In the former, tensile failure was concurrently initiated to give the characteristic tensile and compressive zones on the failure surfaces. In the latter, the growing kink initiated interlaminar cracks in resin-rich zones as it propagated (with decreasing load) towards the convex surface. Kinking was associated with triaxial compressive stresses in the contact zone of the “compressive” roller or rollers (in the case of four-point bend specimens). When hydrostatic pressure was superposed on flexure, at pressures between 150 and 300 MNm−2 depending on the type of specimen, kinking was inhibited and eventually suppressed to give tensile failures, even in the so-called interlaminar shear strength type of specimen. When non-linear deflections were not large, the maximum principal tensile stress in the beams was close to the tensile strength of the carbon-fibre-reinforced plastic (∼1.8 GN m−2).

Cast, wrought, and directly sintered smooth and precracked beam specimens of BT1 steels were studied in three- and four-point bending at room temperature. Following austenitization at 1250° C and tempering between 500 and 560° C, brittle fracture strengths varied between 1.1 and 2.8 GN m−2 and the fracture toughness of the materials was in the range 18 to 25 MN m−3/2. Combining these data, the critical Griffith-Irwin flaw sizes were calculated to be typically of the order of 100 μm. This is in reasonable agreement with the observed sizes of some failure-initiating sites, particularly pores in sintered material, but generally several times larger than the carbide and grain sizes. In wrought specimens, failure frequently originated from groups of carbides, apparently fracturing on contiguous planes. No evidence of sub-critical cracking of carbides was detected (as in BT42), in contrast to BM2, BT6 and sintered and hot isostatically pressed BT1. Only inter-powder particle parting occurred in this sintered material. Conventional fracture mechanics thus successfully interprets results on sintered specimens, but only on several of the wrought specimens. For the majority of the latter it appears necessary to invoke operation of propagation mechanisms involving “short”, ≈10 μm, cracks under monotonic loading or to associate the brittle fracture stress with that for crack nucleation: e.g. cleavage of a carbide cluster.

The deformation and fracture behaviour of rubber-coated and uncoated epoxy specimens has been studied under superposed hydrostatic pressures extending to 300 MN m−2. Maximum shear stress at yield for this epoxy were about 25 MN m−2 at atmospheric pressure and rose to about 48 MN m−2 at 300 MN m−2 superposed pressure. Yielding and failure of all specimens tested beyond pressures of 75 MN m−2 took place when all the (macroscopic) principal stresses, though unequal, were compressive. Fractographic examination revealed three distinct zones of the failure surfaces at atmospheric pressure. The behaviour of all uncoated specimens and those coated and tested below 100 MN m−2 was similar. A fracture-mechanics interpretation of failure could be applied to these tests assuming the deformation-produced first zone was the fracture initiating site. Coated samples tested beyond 100 MN m−2 superposed pressure failed with no evidence of Zones II or III of failure; Zone I appeared to spread over the entire failure surface. An interpretation involving fluid penetration of Zone I failure nuclei, along the lines suggested by Duckett, can account for the failure stresses of the uncoated specimens but is not tenable for the coated samples. It appears that crack nucleation and (slow) growth, as opposed, perhaps, to (catastrophic) crack propagation, can take place in this polymer when all the principal stresses are compressive.

Changes in microstructure in single crystals of MgO, containing elastic discontinuities, which result from pressurization treatments in the range 0.2 to 2 GN m−2 have been studied using an etch-pitting technique. Complex dislocation arrays have been observed around cubical and spherical cavities and precipitates. The observations are discussed in terms of stress and strain criteria. Production of {100} and {110} cracks at high pressure is also described. Following a 1 GN m−2 pressurization the flow stress was observed to decrease by < 10% and this has been related to the induced dislocation density. In polycrystalline MgO the fracture strength was unaltered by pressurization but the critical temperature difference of thermal shock,ΔTC, was increased by 55K. This has similarly been attributed to the movement of pressurization-induced dislocations.

The failure process in waisted tensile specimens of pultruded 60% volume fraction carbon fibre-epoxide was investigated at atmospheric and superposed hydrostatic pressures up to 300 MN m−2. The maximum principal stress at fracture decreased from ~ 2.0 GN m−2 at atmospheric pressure to ~ 1.5 GN m−2 by 200 MN m−2 superposed pressure and then remained approximately constant. These latter failures were fairly flat and no damage preceding the catastrophic fracture was detected, which indicates that composite strength is solely controlled by fibre strength. Fracture of fibres at lower pressures appeared to commence also in the range 1.5 to 1.6 GN m−2, but, as it did not result in catastrophic failure, account has to be taken of the resin and the fibre bundles. Debonding was initiated at ~ 1.2 GN m−2 at atmospheric pressure and this stress increased to ~ 1.5 GN m−2 when 150 MN m−2 superposed pressure was applied; the pressure dependence was related to that of the resin tensile strength. This process is described as the first stage, straightening and debond initiation of curved surface bundles, on our model of tensile failure. The second stage, delamination, i.e. the growth of transverse cracks leading to the detachment of these bundles, was impeded by the transverse pressure, being suppressed beyond 150 MN m−2. Only below this pressure was load redistribution between bundles possible, but, as the pressure was increased from atmospheric, it become more difficult, resulting in a decrease in the composite tensile strength and reduced fibre pull-out.

The failure process in uniaxially-aligned 60% fibre volume fraction glass fibre-epoxide compressive specimens strained parallel to the fibre axis was investigated at atmospheric and superposed hydrostatic pressures up to 300 MN m−2. The atmospheric strength was about 1.15 GN m−2 (about 20% less than the tensile) and strongly pressure dependent, rising to over 2.2 GN m−2 at 300 MM m−2 pressure, i.e. by about 30% per 100 MN m−2 of superposed pressure. The corresponding figure is 22% if the maximum shear stress and not the maximum principal compressive stress is considered. This is incompatible with atmospheric compressive failure mechanisms controlled by weakly dependent or pressure independent processes, e.g. shear of the fibres. The results also could not be satisfactorily interpreted in terms of microbuckling of individual fibres. Kinking, involving buckling of fibre bundles was proposed as the mechanism of failure propagation, but the critical stage (for this glass reinforced plastic) is suggested as being yielding of the matrix, which initially restrains surface bundles from buckling. A strong pressure dependent failure criterion, about 25% increase per 100 MN m−2, was derived by modifying the Swift-Piggott analysis of deformation of initially curved fibres. It is postulated that it is the axial compression that causes bundle curvature. Other systems, particularly carbon fibre-reinforced plastic, in which there appears to be a transition in the critical stage of failure from bundle buckling to matrix yielding with increasing superposed pressure, are also considered.

Single crystals of NaCl, of known surface dislocation substructures and containing precipitate particles of Na2SO4 approximately 10μm in diameter, were subjected to hydrostatic pressures in the range 1 to 10 kbars and subsequently re-etched at atmsopheric pressure. Rearrangement and nucleation of dislocations in the vicinities of the precipitates was observed to result from pressurisation treatments above 3 kbars. After pressurisation at 8 and 10 kbars the dislocation groups were in general complex; linear arrays on 〈110〉 {110} and “rosettes” were, however, observed. The observations are discussed in terms of recent theories (developed for metal matrices) of generation of dislocations at misfitting particles and are in fair agreement with the model of Ashby and Johnson.