Like proteins, starches are natural polymers which can be made into adhesives; however, while proteins form fast-drying liquid glues, starches form slow-drying thixotropic pastes. These pastes can be made by simply heating and stirring a slurry of starch in water until the mixture thickens, and are useful for joining large surfaces to create laminated structures out of porous materials. Unfortunately, they also require a large proportion of water (80-90% by weight) to reach a usable consistency, which tends to saturate and distort the materials being joined during the drying process. However, I have found that this can be avoided by replacing the water in the initial slurry with a suitable low-viscosity adhesive solution. Specifically, I determined that one part of starch (by weight) can be used to thicken four parts of a 25% solution of either dextrin (partially depolymerized starch) or sodium silicate (water-soluble glass), producing a composite paste containing only 60% water and with greatly improved adhesive strength.

   Not all starches are alike however, and neither are all dextrins or sodium silicates. The starch used should ideally be from cereals (e.g. wheat, rice, or corn), since starch from tubers (e.g. potato or cassava) tends to require much more water to achieve an equivalent consistency. Flour may also be used as a crude source of starch, but this contains proteins and other substances which must be taken into account, and is sometimes too coarse to make a smooth paste. Regarding dextrin, it is made using either heat (pyrodextrin) or enzymes (maltodextrin), and both are useful to make pastes; however, maltodextrin tends to be lighter in color and lower in viscosity. Meanwhile, sodium silicate can be classified by its modulus, which is the ratio of silicon dioxide (SiO2) to sodium oxide (Na2O) in its chemical structure. Ideally, an adhesive-grade silicate with a modulus of at least 3 should be used, commonly known as "water glass" and typically sold as a 30-40% solution by weight. This can then be diluted to 25% for use in pastes, while dextrin can simply be dissolved in water to produce the same concentration.

   Whether to use dextrin or sodium silicate depends entirely on the desired properties of the paste. Starch-dextrin pastes are pH-neutral but fully water-reversible, while starch-silicate pastes are alkaline but water-resistant once dry. This resembles the difference between gelatine and casein glue, and in fact these glues are miscible with their respective pastes, producing intermediate adhesives of variable proportions. Interestingly, the alkalinity of silicate pastes also allows for the more efficient utilization of vegetable proteins. Specifically, if flour containing gluten (e.g. from wheat, rye, or barley) is used rather than pure starch, the gluten fully dissolves and becomes an auxiliary adhesive similar to casein, resulting in a more durable bond. This also allows the paste to be used "in reverse", by using a flour-silicate slurry as a temporary adhesive between two surfaces, then applying heat and pressure to thicken the mixture in-situ. Care should be taken though, to ensure that the heat does not boil the paste, and that the pressure does not squeeze it out of the joint. This reverse method also allows for twice the usual amount of flour to be used (two parts flour to four parts dextrin or silicate solution), which further reduces the water content of the paste to 50% by weight, comparable to the best gelatine glues. As a side note, I have noticed that these silicate pastes have no tendency to form efflorescent deposits of sodium carbonate, unlike sodium silicate when used alone. This is likely due to the proteins and starch absorbing the excess sodium produced during the curing process, which appears to be a useful principle when working with silicates in general.

   The many variations of this paste can be seen in use at the top of the page, bonding a variety of materials to a base of high-density fiberboard. The first two materials (paper and cotton) are bonded with a paste of corn starch and dextrin. The next two (wool and leather) are bonded similarly, but with a paste of wheat flour and sodium silicate. The final material (mahogany veneer) uses a hot-pressed slurry of rye flour and sodium silicate, in which the dissolved proteins and soluble carbohydrates from the flour serve to suspend its own starch. This last process is of particular interest, as I believe it could be used to produce plywood from simple materials at very low cost. The pastes used to make these samples, as well as a closer view of the layered structure of the samples themselves, can be seen below.

   These examples are by no means comprehensive, however. Adjustments can be made as needed to the ratios and concentrations of the ingredients of the paste, in order to control its viscosity and adhesive properties. Preservatives may also be added such as clove or thyme oil, though these are strictly useful for dextrin pastes, as silicate pastes are too alkaline to support microbial life. Plasticizers such as glycerine or sugar can also be added to dextrin pastes to improve their flexibility and adhesion, at the proportional expense of increased water sensitivity. Interestingly, sugar often occurs naturally in dextrin as an impurity, since glucose is the monomer of starch (and therefore, the end product of depolymerization). Silicate pastes on the other hand are fundamentally rigid; however, their elasticity can be improved somewhat by the incorporation of proteins such as casein or gluten. Finally, fillers such as wood flour or mineral powders may be added to either type of paste, if a gap-filling material is needed for joining rough surfaces or patching small holes.

   Aside from the ingredients themselves, application methods may also vary. Some materials (e.g. cotton or wool) are too porous to apply paste to directly and must be stuck to the wet surface of the substrate, while other materials (e.g. leather or wood) may be coated on both faces, or even just on the material itself (e.g. paper). Furthermore, a wide variety of equipment can be used to apply pastes, depending on the nature of the adhesive. Thick pastes may be applied with a trowel or spreader, and thin pastes with a brush or roller, with a brayer or rolling pin being useful in all cases to ensure solid contact between the materials being joined. Meanwhile, hot-pressed pastes can be set using a clothes-iron for small-scale work, but also using a veneer press, as in the industrial production of plywood. Overall, the adjustability of this adhesive makes it useful for a wide variety of tasks, ranging from paper crafts to the manufacture of construction materials, and I believe the formulas I have developed are inexpensive and strong enough to make this feasible on a large scale. I look forward to using this type of paste in upcoming projects, and to the creative and structural possibilities it brings to my future work.