Anonymized use case. Client, company and supplier names are withheld. This is screening-grade analysis built from published material properties, supplier documentation and food-science literature. None of it is bench-validated in the client’s process.
The problem
A white soy-protein piece, about 50% protein, needed to lose its titanium dioxide. It is mixed in, slightly acidified before packing, shelf-stable for 12 months, and it goes through a hard thermal path: steam extrusion near 145°C, then a long dry near 125°C. The piece already shows a light off-color.
The first screen: acid and heat
DyeConverter™ ranked the white colorants on the two things this process punishes, acid and heat:
- Leads: tricalcium and calcium phosphate whites (acid-stable, heat-stable, and they add calcium and phosphorus), and a structural cellulose white (inert to acid and heat, strongest scattering per gram of the group).
- Fallback: nano-free calcium sulfate. It does not fizz in weak acid; that behavior belongs to carbonate.
- Ruled out: calcium carbonate, which fizzes and neutralizes the acid step, and native starch, which gelatinizes and loses opacity in the extrusion heat.
Why the mineral screen plateaued
On the bench, the leads gave too little lift. The reason is optical. Whiteness comes from light scattering, and scattering depends on the refractive index (RI) contrast between the particle and the material around it, not on how white the powder looks in the bag.
| Material | RI | Contrast in water (1.33) | Contrast in dense protein (~1.55) |
|---|---|---|---|
| Titanium dioxide (rutile) | 2.70 | 2.03 | 1.74 |
| Calcium phosphate | ~1.64 | 1.23 | 1.06 |
| Calcium carbonate | ~1.57 | 1.18 | 1.01 |
| Calcium sulfate | ~1.52 | 1.14 | 0.98 |
| Cellulose | ~1.47 to 1.55 | 1.13 | ~1.00 |
That explains the cellulose result in particular. In a beverage it sits in water and scatters well. Inside a dense protein piece it is nearly index-matched and close to invisible. That is a matrix mismatch, not a bad ingredient. Particle size is the second axis: scattering peaks near 0.2 to 0.3 microns, which is exactly where titanium dioxide pigment is milled, and many mineral whites ship far coarser.
The highest-leverage move: lower the bar before raising the paint
If the off-color is reduced at the source, the whitening job shrinks, and options that almost worked may become enough. These levers cost nothing on the label.
- The soy protein grade. A piece near 50% protein is typically built from defatted soy flour, the tannest starting material, which keeps its soluble sugars, the Maillard fuel entering a 145°C extruder. Alcohol-washed concentrate is lighter with those sugars largely removed; isolate is lighter still. Moving up that ladder whitens the base and removes browning precursors in one step.
- pH timing. Maillard browning runs fastest around pH 6 to 8. Acidifying earlier, or buffering lower during the thermal steps, slows it while it is forming.
- Reducing sugars and protein reactivity. Fewer free reducing sugars, or a less reactive protein cut, starve the reaction.
Re-test what you already have
Phosphate was not incompatible; it was underpowered alone. Titanium dioxide doses near 0.1 to 1%, and mineral whites need several times that. In soy protein, calcium salts also act on the protein (calcium sulfate is the traditional tofu coagulant), so the trials may have hit a texture or chalkiness ceiling before a whiteness ceiling. If dosing was capped for mouthfeel, the result reflects the cap, not the material.
Non-mineral options, in parallel
- A proprietary non-mineral whitener marketed for protein foods and plant-based meat, claimed heat- and acid-stable. Composition is proprietary, so it has to be proven in the process, not assumed.
- Engineered emulsion whiteness, the mechanism milk uses to look white without titanium dioxide. It is documented in meat analogs, so it works in a solid protein matrix, but it reads creamy rather than bright.
Honest exclusions
- Rice, corn and tapioca starch opacifiers gelatinize around 65 to 80°C with moisture present, well below a 145°C wet extrusion.
- Rice-protein-coated whites built on a calcium carbonate core carry both the mineral ceiling and the acid reactivity.
- Water-soluble whites work against opacity in a solid piece.
- Protein-based photonic whiteners are promising but pre-commercial: tracking, not recommending.
The bottom line
There is no drop-in non-mineral powder that matches titanium dioxide alone in a hot, wet-extruded solid. The realistic path is combined: cut the browning at the source so less opacity is needed, confirm whether the mineral trials hit a dose ceiling rather than a performance ceiling, and evaluate a non-mineral contributor in parallel. That is a narrower, cheaper bench plan than screening every white on the market.
Replacing titanium dioxide in a tough process? Talk to us.
Frequently Asked Questions
Why do titanium dioxide replacements look weaker?
Whiteness comes from light scattering, which depends on the refractive index contrast between the particle and its surroundings. Titanium dioxide sits near 2.7. Food-grade mineral and cellulose whites sit near 1.5 to 1.7, so they scatter far less.
Why does cellulose whiten a beverage but not a protein piece?
In water (refractive index 1.33) cellulose scatters well. Inside a dense protein matrix (about 1.55) it is nearly index-matched, so it becomes close to optically invisible.
Can calcium carbonate replace titanium dioxide in an acidified product?
No. Carbonate reacts in acid: it fizzes, gives off carbon dioxide and neutralizes the preservation acid. Phosphates and calcium sulfate do not.