Methods of formaldehyde analysis include the iodometric, sulfite [31], and mercurimetric [32,33] methods. The sulfite method measures only the formaldehyde present, whereas the iodometric method can also estimate the methylol groups. Another method is based on the partition of formaldehyde between water and isoamyl alcohol [34]. Estimation of the formaldehyde in the alcohol phase of a mixture of an aqueous solution of the resin and isoamyl alcohol allows deduction of the amount of free formaldehyde. This procedure has the advantage that no risk of reaction arises between free formaldehyde and the resin components.
Kappelmeier [35] has suggested the use of aniline, benzylamine, and phenyl-ethyl — amine as reagents for the identification and analysis of urea in UF resins. He has provided evidence that the methylene-ether groups form a bridge between urea residues in UF resins. The use of benzylamine in particular (which yields dibenzylurea from urea derivatives), has been developed as a method of analysis. In determining the ratio of urea to formaldehyde in UF resins, the benzylamine method has been coupled with a process of formaldehyde estimation which involves depolymerization with phosphoric acid, followed by distillation into alkaline potassium cyanide solution [36].
Chow and Steiner [14] advocate the use of bromination in CCl4 and subsequent x-ray analysis to determine available reactive methylol groups. High-resolution nuclear magnetic resonance (NMR) has also been used to analyze UF resins and to trace their kinetic behavior [37]. Particularly useful is 13C-NMR analysis of liquid UF resins, where clear identification of monomeric species, methylolated or not, methylolureas, methylol groups on the polymer, methylene-ether linkages, methylene bridges, sites of branching, uron, free formaldehyde, and other features can be achieved easily and rapidly [26,38,39]. For example, this technique makes it possible to easily estimate the probable bonding ability and approximate emission class to which the bonded boards are likely to belong [24,26].