6 TEST RESULTS

The results of the shear and tensile tests are reported in tables 2 and 3 respectively. The shear and tensile tests were carried out at Allied Signal FM&T using applied tensile load at a rate of 0.1 in./min. (crosshead speed) at 20.5�C (69�F), +/− 2�, and an RH range of 50-55%. In both shear and tension loads, the epoxy and polyester adhesive bonds were stronger than the marble. Under shear stresses the polyester assemblies failed within the marble at an average stress of 4,101 kPa (595 psi), while the epoxy-bonded assemblies failed within the marble at an average stress of 4,742 kPa (688 psi). In the tensile evaluations the epoxy bonded assemblies failed within the marble at an average stress of 4,519 kPa (655 psi), and the polyester-bonded assemblies failed, at an average stress of 4,616 kPa (669 psi). The results are not unexpected, since the strength of both adhesive types is quite high and was assumed to be higher than that of the two marble types making up the coupons.

It is interesting that the shear coupon bonded with epoxy with added fumed silica showed a significantly higher fracture strength, 15,474 kPa (2,244 psi), than did the solid marble coupon or the coupons bonded with either unbulked epoxy or polyester tested under shear load. Although this coupon was initially prepared to evaluate the bulked epoxy as an adhesive or gap filler, the results suggest that fumed silica may have significant effects on the physical characteristics of an adhesive. While some investigations have been reported on this topic in the conservation literature (Byrne 1984), more work clearly needs to be pursued.

The explanation for the higher bond strength of this bulked sample as compared to the marble control sample is unclear; however, improved stress distribution may be part of the answer. Brittle materials like marble fail catastrophically rather than progressively like malleable materials such as metals. Failure in such brittle materials occurs when the stress anywhere exceeds the maximum that the material can withstand. If the stress distribution is nonuniform, failure still occurs when the highest stress anywhere in the joint reaches the failure stress limit. For a nonuniform stress distribution, the average stress computed will be lower than the actual maximum stress that precipitated the failure since some of the joint was below the failure stress. This relationship implies that in most cases the reported failure or average stress will be lower than the actual failure or local stress, due to nonuniform distribution of the stress. It also implies that any change in bonding methods that improves the uniformity of stress over the joint will enable the joint to carry a higher load, and the average stress at failure will be higher, even though the actual, local, failure stress in the marble is unchanged. Thus it can be postulated that the bulked epoxy distributed the load more evenly over the surface of the bond and hence increased the overall effective strength. It may appear that the marble broke at considerably higher stress levels, higher than the value for whole marble, but what actually occurred is that the bulked epoxy bond is so strong that it distributes the stress more evenly to the marble. The marble then seems to break at much higher stress levels.4

The tests using 1:1 w/w solutions of B-72 used as an adhesive provided interesting data as well. In shear the B-72–acetone mixture failed within the bond line at an average stress of 1,475 kPa (214 psi). However, in tension the B-72–acetone mixture failed at an average stress of 4,545 kPa (659 psi) and the B-72–toluene mixture at 4,316 kPa (626 psi); both failed in the marble, not in or at the bond line.

When B-72 was used as a barrier coat, there was again surprising information from the test results. In shear the B-72–acetone barrier combined with the epoxy bond failed at an average stress of 2,669 kPa (387 psi). All five samples in this case failed in the marble parallel to the bond line in shear failure. In tension, the B-72–acetone barrier combined with an epoxy bond failed at an average stress of 4,399 kPa (638 psi) and the B-72–toluene barrier combined with an epoxy bond failed at an average stress of 4,426 kPa (642 psi). Although the differences between the shear and tensile results are significant, it is worth pointing out that in tensile all the coupons, and in shear most of the coupons, bonded with a B-72 barrier failed in the marble and not within or at the bond line.

When polyester was used as the interlayer adhesive between the B-72 barrier layers, shear failure occurred within the marble parallel to the adhesive line at an average stress load of 3,350 kPa (486 psi). In tension the failure again occurred within the marble, often far from the bond line, at an average stress of 4,472 kPa (649 psi) for the B-72–acetone barrier layers and 4,688 kPa (680 psi) for the B-72–toluene barrier layers, both with polyester bond lines.

The results of the reversibility tests indicated that the bonds made with B-72 as an adhesive and those made with B-72 as a barrier coat were considerably more reversible than those made with only Araldite AY 103 epoxy or Akemi polyester. All the bonds made with B-72 as an adhesive or as a barrier reversed within the first 8 hours of solvent fuming. The polyester bond reversed within 16 hours. After two months in the fume chamber (at which time the test was concluded), the epoxy bond remained intact.