Why fibers and fiber composites are strong, and why bulk materials usually aren’t.
One may wonder why soft armor materials — and other materials for defense and aerospace applications that exhibit a very high strength-to-weight ratio — all happen to be fiber composites. Why aren’t they bulk materials? Why is graphitic fiber so much stronger than graphite, why is fiberglass so much stronger and tougher than bulk glass, and why is it that basalt fibers are now heralded as a new-age superfiber, whereas basalt itself is almost completely useless? The below account of the origins of fiberglass may prove both interesting and enlightening. It is taken from the excellent “The New Science of Strong Materials,” by James Edward Gordon, one of the founders of materials science as a science:
“[Alan Arnold] Griffith had first to determine, at least approximately, the theoretical strength of the glass he was using. The Young’s modulus was easily found by a simple mechanical experiment and two or three Ångström units is a fair guess for the interatomic spacing and cannot be far out. It remained to measure the surface energy. It was here that one of the advantages of glass as an experimental material lay. Glass, like toffee, has no sharp melting point but changes gradually, as it is heated, from a brittle solid to a viscous liquid and during this process there is no important change of molecular structure. For this reason one might expect there to be no large change in surface energy between liquid and solid glass so that surface tension and therefore surface energy, measured quite easily on molten glass, ought to be approximately applicable to the same glass when hardened. When the end of a glass rod is heated in a flame the glass softens and tends to round off into a blob because surface tension remains active long after permanent mechanical resistance to deformation has disappeared. The force, which is easily measured, needed slowly to extend the rod under these conditions is therefore that which will just overcome the surface tension. From experiments of this type, done with very simple apparatus, Griffith could deduce that the strength of the glass he was using (at room temperature) ought to be nearly 2,000,000 p.s.i. or about 14,000 MN/m2. [MPa]
A large portion of Professor Gordon’s book is devoted to this question, which boils down to microstructural defects, atomic inclusions, inhomogeneities, and other flaws that are always present in bulk materials but which are reduced — or hardly present at all — in very thin fibers or in nanomaterials such as graphene.It is impossible to prepare a bulk ceramic material without flaws, but if such a thing were possible, that ceramic would be both incredibly strong and incredibly tough. It would seem like an entirely new class of material. Sadly, again, this is effectively impossible. Consider — when water freezes, ice never forms perfectly uniformly and regularly. This is why, in a cube of ice you would drop in your drink, you will always find small defects: Bubbles and lines and planes of fracture. Ceramics, metals, etc. are much the same. Some materials, like ductile metals, can compensate for defects to some extent; other materials, like ceramics and glass, cannot. What’s clear is that all bulk materials are imperfect. Indeed, the universe itself is not perfectly homogenous, but contains clumps of matter, voids, and, perhaps, even stranger defects.In any case, it’s worth noting that the strength or stiffness of a composite material is not equal to the strength or stiffness of its fiber component. For example, although the tensile modulus of an average high-quality UHMWPE fiber strand should be somewhere around 115 GPa, the tensile modulus of a sheet of Dyneema is always far lower. The modulus of the composite can be approximated, very roughly, via a simple calculation:E1 = κEf1Vf + EmVm
Where Ef, Em, Vf, and Vm are the moduli and volume fractions of the fiber and the matrix, respectively, and where κ is a factor that concerns bonding behavior at material interfaces under axial strain. In most relevant cases, κ = 0.919.
There are better models, but they are considerably more complex. The above is presented merely for illustrative purposes.