The various silicon powders listed in Table  1 were dissolved/dispersed in drinking water (The National Forest Vending Co. Ltd., Newthorpe, Nottingham, UK, pH 6.8 at source) at different concentrations. The taste solutions of metallic salts (ultrapure grade) and silicon-based powders were prepared in drinking water according to International Standards Organization guidelines 3972:1991 using the purest salts available and the powders listed in Table  1.

The screening tests were first performed with the purpose of defining individual threshold concentrations that can be detected. Volunteers were presented with cups coded with random numbers; each cup contained a 20 ml test solution (in varying concentrations). When a subject perceived a distinct taste sensation, he or she recorded and described the particular taste. Once the individual’s threshold was defined, in the next session the subject underwent a modified triangle test. The triangle test is a three-product test in which the task is to identify one sample that differs from the other two. Three containers of identical volume, covered with aluminium foil, were presented to volunteers and they were asked to identify the odd sample and indicate the perceived taste. The samples were coded in a random order (e.g. CCA, ABA and ABC) with one sample being the test and the other two being pure water. The major impurities of the drinking water are listed in Table  2.

Si, OpSi and pSiO₂-loaded chewing gum pellets were made by initially vortex mixing a 27 g batch of chewing gum base powder (Cafosa Gum Ltd., Barcelona, Spain) with 3 g of the test substance. Aliquots of 1 g were then cold pressed into pellets using a 5-mm die set with 10 kN force for 30 s.

The chew-out tests for metallurgical grade silicon and mesoporous silicon particles were done by a 10-person panel. The volunteers were asked to chew the pellet for 2 min. Two minute chews were followed by the collection of the saliva and rinse water for analysis of loss of silicon particles during chewing and recording the observations. The volunteers were asked to grade the samples for grittiness, taste and aftertaste.

Chew-out tests were also mimicked in vitro by mechanically grinding pellets in Tris buffer at pH 6.8/artificial saliva for 0, 2 and 10 min. The release of silicic acid in media was measured at different time points and compared with silicic acid release from pellets without grinding. The media were filtered and analysed for silicic acid content.

The taste of a silicic acid solution of 58 μg/ml in deionized water was also judged using a triangle test. Mesoporous silica (Davisil, pore diameter 15 nm, particle size 52.5 μm (D50)) was completely dissolved in Tris buffer over a 12-day period. The samples from this solution were analysed for determination of the silicic acid released in the media. From the same sample, 6 ml aliquots were presented to volunteers in sets of three, where only one sample was the test and other two were deionised water used in preparation of buffer. The volunteers underwent the triangle test again as described earlier.

The silicic acid content was measured by spectrophotometry using molybdenum blue assay 13 . It is based on the reaction of silicic acid [Si(OH)₄ with molybdic acid (or ammonium molybdate) at pH 1.5 to 2 to form the yellow isomer beta silicomolybdate (SiMO₁₂O₄₀)⁴⁻. This molybdate complex is then reduced by sodium disulphite to silicomolybdenum blue to increase spectrophotometric sensitivity. This allows determination of silicic acid concentration in the range of 10 to 70 μgs/ml with a variance of ±10%.

The mouthfeel of particles is very important in food products as it affects mastication and overall taste sensation of the product. In our study we chose particles (silicon/silica) of varied sizes to study the grittiness and any possible metallic taste or aftertaste of chewing gum pellets. All the particles used were analysed for particle size distribution and BET analysis for measuring porosity or pore volume.

Figures  1 and 2 illustrate the particle size distribution of two divergent samples from the six different particle sizes used in the study. Solid silicon powder with the smallest particles had a (D50) particle size of 4.4 μm whereas the largest mesoporous silica particles had a mean (D50) particle size of 134 μm.

The taste of different types of silica/silicon powder was compared by using batches of similar particle size; namely, solid silicon (D50 of 4 μm), OpSi powders (D50 of 4 μm) and pSiO₂ particles of (D50 of 11 μm). The data shown in Table  3 indicate the concentrations of solutions which the panel could detect significantly when compared with pure drinking water. Almost 98% of volunteers could detect copper sulphate and solid silicon at 0.2 and 10 mg/ml respectively; therefore, the level of significance of the test was the highest (p < 0.001). Oxidized porous silicon could be tasted at 1 mg/ml by 85% of the panel (p < 0.01), whereas only 58% of the panel could detect porous silica in water at the same concentration and therefore, the level of significance for the latter is only p < 0.2.

With regard to taste descriptors, ‘supertasters’ within the panel of 22 volunteers ascribed a ‘chalky’ taste to silica in water and ‘metallic taste’ to bulk silicon. Their taste descriptors for oxidized porous silicon were ‘no metallic taste’ or ‘no off-taste’.

The pellets made from various silicon/silica materials are shown in Figure  3. Those made with solid silicon are black (Figure  3a), those with oxidised porous silicon are brown (Figure  3b) and those with porous silica were off-white (Figure  3c). Since we wanted to explore at which size the grittiness of silicon/silica particles in chewing gum pellets could be detected, a wide range of the particle size (4 to 134 μm) was chosen. The comparison of mouthfeel for bulk and porous silicon utilized similar particle size distribution.

The chewing pellets prepared all weighed 1 g and they were presented to the volunteers in a blind manner to avoid biased taste results. Chew-out tests from gum samples showed varying mouthfeel for different material groups. For 10 wt.% OpSi-loaded gum, it would appear that a D90 of <12 μm would be acceptable, and 54% of volunteers found no ‘off-taste’ or ‘aftertaste’ at this high loading. The 10 wt.% pSiO₂ gums, prepared from either particle size distribution, consistently had a ‘chalky’ taste. Such a high loading of particles had a significant effect on the mechanical properties of the pellets for all the test samples.

The in vitro study of biodegradation of the OpSi particles (D50 of 4 μm) in mechanically grounded gum (Figure  4) found a measurable level of silicic acid for storage times of 2 and 10 min. In contrast, pellets immersed in artificial saliva/Tris buffer pH 6.8 without grinding exhibited a negligible release of silicic acid (Figure  4), indicating the importance of mastication forces for accelerating the degradation of silicon/silica particles.

The threshold detection level for silicic acid in water was not quantified. Nevertheless, selected members of the panel were clearly able to identify highly concentrated silicic acid solutions from water controls with 100% accuracy. Their taste descriptors for the 58 μg/ml orthosilicic acid solution were ‘bitter sweet’ or ‘sweet’.

The primary objective of the study was to gauge the perceived taste of the semiconductor silicon, in solid and porous microparticulate form. The semiconducting silicon exposed to air or an aqueous environment will be covered in an ultrathin native oxide, so one might conclude that it is silica rather than silicon that is being assessed by the taste buds of the tongue. This becomes even more relevant for silicon structures that have been intentionally thermally oxidised (refer to Table  1).

However, there was a clear disparity between the panel perception of the taste of the solid silicon and the silica structures. This suggests that the ultrathin native oxide of silicon might not be effective in protecting taste bud exposure to the underlying silicon in the corrosive mouth environment. This is not too surprising, considering that much thicker plasma-enhanced chemical vapour deposition silica films on silicon were found to continuously corrode in vivo 14 .

In the drinking water tests the particles are exposed to diluted human saliva for only a very short period of time (30 s). In the chew-out test they are exposed to both mastication forces and diluted human saliva for longer periods (2 min). Microparticle breakage exposes fresh surfaces to saliva and to the tongue. In addition, the biodegradability of mesoporous silicon is now established 8 9 , so one would expect more degradation of the mechanically weaker porous silicon particles in the mouth, exposing material underneath that of the ultrathin oxide coating. Prior in vitro studies (QS and A Pokale, unpublished work) have shown that the rate of biodegradation is lowered as a result of thermal oxidation over the 300°C to 800°C temperature range. Significant silicic acid release still occurs after 600°C oxidation, even in the absence of mastication forces, but over many days of storage in simulated body fluids.

The data of Figure  4 (upper trace) support the statement that the oxidised porous silicon particles underwent significant biodegradation under the chew-out test conditions. However, the Figure 4 data (lower trace) also suggest that for the water test conditions, a very low degree of particle degradation occurred. It is also unclear as to whether the levels of silicic acid released into the mouth by either test would themselves have imparted a significant taste.