A butterfly pea extract that looks convincingly blue in water has not yet qualified for a blue protein drink. Two decisions still sit between the sample and the product: whether the required acidity allows the intended shade, and what happens when the extract meets the actual protein, sugar and mineral system. Published experiments make these separate problems unusually clear. A spectral study described butterfly pea extracts as bright blue at pH 5–8, whereas a whey-protein study showed that changing the surrounding sugars and salts changed pigment–protein binding.[1][7]
The useful purchasing question is therefore not simply “How strong is your blue?” It is “Can this material give our required blue in this base, through this process?” A stronger extract cannot be assumed to solve a mismatch between shade and pH; a protein–pigment interaction cannot be assumed to provide shelf-life protection.
Blue is a state of the pigment, not a fixed property of the powder
Butterfly pea owes its blue colour to ternatins, a group of polyacylated, delphinidin-based anthocyanins.[1][7] They are not interchangeable with every other material sold as an anthocyanin colour. In the 2021 spectral study, the extract was magenta from pH 0.5 to 3, shifted towards purple between pH 3 and 4, and was described as bright blue from pH 5 to 8.[1] At higher pH, the spectra changed again: the authors linked the growing short-wavelength absorption above pH 8 to yellow chalcone formation, with blue-green and ultimately yellow appearances across the alkaline range.[1]
There is an important qualification inside this apparently convenient colour chart. Those samples were extracted with 1% HCl–methanol, adjusted with sodium hydroxide and examined as laboratory solutions—not as a commercial water-extracted colour in a finished drink.[1] The results identify a direction for screening; their pH boundaries are not a guaranteed shade specification for every supplier’s product. Even the study photograph shows a shift towards cyan at pH 8, rather than one identical blue throughout the range.[1]
The proposed structural explanation is useful nonetheless. The long acylated side chains of ternatins can favour intramolecular stacking, protecting the pigment against hydration and helping preserve blue quinonoid forms.[1] The authors explicitly described the detailed pH-dependent structures as speculated from the colour and spectra.[1] It is a mechanistic interpretation, not a direct measurement of the molecular population inside a milk drink.
For an acid-led fruit/protein beverage, this creates a real formulation choice. If the base must remain strongly acidic, first ask whether a purple or magenta shade is acceptable. Do not begin by increasing the blue-colour dose and only later discover that the target hue is incompatible with the base. Conversely, a milder-pH blue concept should be developed within the product’s independently validated safety and process design; raising pH to improve a photograph is not a preservation strategy.
The neutral-pH result is promising—but does not mean “heat stable”
In the same study, the blue extract at pH 7 retained its measured absorbance better than the magenta pH 0.5 and blue-green pH 10 samples under the tested storage conditions.[1] That runs against the simple purchasing rule that anthocyanin colours should always be made more acidic. It does not mean butterfly pea tolerates any heat exposure.
At pH 7, the reported loss in absorbance at 576 nm after 27 days in darkness was 4.7% at 4°C, 32.8% at 25°C, 51.7% at 37°C and 63.6% at 50°C.[1] These are losses in a specified spectral peak in the acidified-methanol-derived test system, not finished-beverage colour rejection rates or pigment mass losses.[1] Even the favourable pH did not remove the temperature effect.
The paper’s fitted half-life of 334.2 days at pH 7 and 4°C can also be easily oversold.[1] Quantitative absorbance measurements ran to day 27; the study separately photographed colour after 65 days.[1] A model-derived half-life is not 334 days of observed commercial shelf life, and neither dataset validates a short high-temperature process. For procurement, request the supplier’s actual test matrix, exposure time, package and measurement wavelength alongside any “heat-stable” claim.
What whey protein experiments actually establish
A second 2021 study used a more application-relevant extraction route: dried butterfly pea flowers were extracted in water at 80°C for 30 minutes, at a flower-to-water ratio of 1:100 by weight.[7] Those are the researchers’ preparation conditions, not an extraction yield optimum or a factory recipe.
The investigators then studied commercial whey protein isolate (WPI) with the flower extract. They used fluorescence quenching to examine interactions: as more extract was added, the protein’s fluorescence declined, and the resulting measurements were used to estimate binding constants and binding-site numbers.[7] Pigment concentrations were expressed as ternatin B2 equivalents because ternatin B2 was the major identified anthocyanin in their extract; this was not a trial with pure isolated ternatin B2.[7]
The distinction matters. The study supports interaction between a mixed flower extract and a mixed whey-protein material. It does not establish that an arbitrary commercial powder will behave identically, or that one constituent alone caused every measured response.
The baseline WPI solution prepared in distilled water had a reported pH of 5.5.[7] In the initial binding experiments, final protein concentration was 5 mg/mL; in the sugar and calcium experiments it was 2.5 mg/mL.[7] Samples were incubated for 30 minutes at 35, 45 or 65°C before fluorescence measurements.[7] These details define a mechanistic experiment, not a high-protein ready-to-drink formulation or an industrial sterilisation challenge.
The authors inferred spontaneous, exothermic binding with hydrogen bonds as the principal force in the distilled-water system.[7] Increasing temperature from 35 to 65°C reduced the calculated binding constant and number of binding sites.[7] A buyer should therefore resist the appealing shorthand “protein encapsulates the colour, so heating helps.” This particular experiment found weaker binding as temperature increased, and did not demonstrate improved beverage shelf life.
Sugar and minerals are part of the colour system
The same paper offers a useful negative result. At 35°C across the tested 0–150 mM added-sugar range, Figure 4 shows no significant concentration effect on the binding constant for either lactose or sucrose, or on the fitted binding-site number for sucrose.[7] Some lactose concentrations did differ in the estimated number of accessible sites.[7] Thus, replacing one sugar system with another is not adequately described as changing sweetness while leaving the pigment environment untouched—but adding sugar is not automatically an effective binding treatment either.
The calcium result requires even more care. At 35°C, Table 3 reports binding constants of (13.56 ± 4.73) × 10⁴ M⁻¹ without added calcium chloride, (20.63 ± 1.36) × 10⁴ M⁻¹ at 0.375 M and (26.32 ± 1.94) × 10⁴ M⁻¹ at 0.500 M. These are means ± SD from three separate trials; the rows carry different significance letters.[7] The abstract and discussion describe the increase as occurring “above 375 mM,” although the table already marks the 375 mM row as different from the lower-concentration group.[7] The safe reading is that the stronger-binding effect appeared at the high end of this experimental series, not that the paper establishes a precise formulation threshold.
Those concentrations are not calcium-fortification recommendations. More importantly, binding affinity was the measured endpoint: the experiment did not establish acceptable flavour, colloidal stability, permitted addition levels or longer shelf life at those salt concentrations. Ask for performance at the mineral composition your product actually needs, rather than adjusting a drink to reproduce the highest binding constant.
The paper also reports that its hydrogen-bond interpretation differed from the authors’ earlier buffered-system work, which had identified hydrophobic interactions as the main force.[7] This comparison comes from the later paper’s discussion of the earlier work. The practical lesson is narrower and stronger: buffer chemistry, salts and non-protein constituents belong in the transfer assessment. “Whey protein present” is not a complete specification.
A useful blue-drink comparison changes one decision at a time
The following is a development approach drawn from those findings, rather than a validated formulation protocol.
First settle the shade in the base. Compare the proposed colour in the uncoloured drink matrix across the product’s feasible pH range. Agree whether the target is blue, blue-violet or purple before discussing a higher dose. Record the pH after the full base is assembled, not only in the colour supplier’s reference solution.
Compare protein grades while holding the other formulation conditions constant. When comparing WPI grades, retain the same target protein concentration and document the lactose, ash and added salts. Include a colour-only control and a protein-base blank. This makes it easier to distinguish a pigment response from a change in the underlying appearance of the base. It does not make one WPI trial evidence for casein, pea protein or soy protein.
Keep a real-world arm alongside the simplified comparison. The finished sugar/mineral blend may be the commercial requirement, even if a simpler laboratory solution gives stronger binding. Test that real blend, including its intended addition sequence and process, rather than optimising only the model system.
Judge the finished appearance after processing and storage. Record shade and colour difference against a reference, and separately examine haze, sediment or separation. Use an agreed optical method for the actual matrix; do not convert fluorescence quenching into a colour-retention claim. A colour that interacts strongly with protein still has to meet the drink’s visual and physical requirements.
This approach also improves the quotation. Ask suppliers to identify the botanical material and plant part, extraction route, liquid or powder format, carriers and standardisation method. Request the basis of any anthocyanin result—reference compound, method and reporting units—and a colour-strength test with its pH, concentration and optical conditions. Compare cost at the accepted shade in the actual drink, not price per kilogram alone.
Permission and performance are different gates
For a US project, the FDA’s May 2025 final order describes the relevant butterfly pea flower extract as a dark-blue liquid produced by water extraction of dried Clitoria ternatea petals.[8] The order also notes that the aqueous extract must undergo ultrafiltration under 21 CFR 73.69(a)(1).[8] It lists uses including nutritional beverages and dairy and non-dairy drinks at levels consistent with good manufacturing practice (GMP), while retaining the restriction for foods with standards of identity unless those standards authorise added colour.[8] Its wording is specifically “fruit preparations in yogurts”, not an unrestricted statement about every yogurt application.[8]
The 2025 change expanded uses to specified cereals and snacks; a later notice confirmed June 26, 2025 as the effective date.[8][9] These official documents do not qualify every dried-flower powder, laboratory solvent extract or protein complex for every use. Match the supplied material and the intended food category to the applicable requirements before commercialising. This discussion establishes neither China nor EU market access.
For the colour brief itself, the conclusion is straightforward: choose butterfly pea for the shade it can deliver in the required base, then investigate protein interactions as a formulation variable. The credible opportunity is not a universal blue replacement. It is a blue or blue-violet system whose pH, protein grade and dissolved ingredients have been considered together—and whose finished-product performance has actually been measured.
Related reading
- Beverage applications
- Research library
- Hibiscus colour and anthocyanin endpoints
- Beet colour in plant-based meat
Sources
[8] FDA (2025-05-12). Listing of Color Additives Exempt From Certification; Butterfly Pea Flower Extract
[9] FDA (2025-08-21). Confirmation of Effective Date: Butterfly Pea Flower Extract