You bench a clean blue hard candy. You drop a few crystals of citric acid into the mix to land the sour profile. Before the syrup has even cooled, the blue is gone. In its place: vivid purple. Sometimes red.
This is the classic butterfly pea flower failure. It happens in candies, in lemonades, in cocktails, in gummies, in carbonated water. It happens every time someone treats butterfly pea like a drop-in replacement for FD&C Blue 1.
Butterfly pea is not Blue 1. It is a fundamentally different chemistry, with a different stability profile and a different set of rules. The good news is the rules are knowable, and the lab failure is solvable.
What Is Actually Happening Inside the Solution
The blue pigment in butterfly pea flower (Clitoria ternatea) is technically an anthocyanin, but lumping it in with "anthocyanins" misses what makes it useful. The specific molecules are ternatins: a series of 15 polyacylated delphinidin glucosides first structurally characterized by Terahara et al. between 1990 and 1998.
The backbone is delphinidin 3-O-(6"-O-malonyl)-beta-glucoside, decorated with alternating D-glucose and p-coumaric acid side chains at the 3' and 5' positions. Those acyl groups matter. They stack with the chromophore in an intramolecular sandwich and stabilize the blue-favoring quinoidal form. Strip the acylation away and you no longer get blue. White-flowered Clitoria ternatea lines lack the 3' and 5' glucosylation entirely and never go blue, which is the cleanest possible proof that polyacylation is what creates the blue (Kogawa et al.).
That structural detail is why ternatins behave better than other natural anthocyanins. It is also why they still cannot survive a pH crash.
The Real pH Band -- Not the One in the Food-Industry Blogs
A lot of online sources, including some of my own earlier social posts, repeat the line that butterfly pea is blue from pH 3.8 to 7. That range is off. Here is what the actual spectroscopy (Jiang et al. 2021; Marpaung et al. 2017; Jeyaraj et al. 2024) reports:
- pH 0 to 2: red to magenta. The flavylium cation form dominates. Absorbance peak around 548 nm.
- pH 3 to 5: red-purple shifting to purple as pH rises. Mixed flavylium and quinoidal forms.
- pH 5 to 7: blue. Quinoidal base plus intramolecularly copigmented species. Absorbance peak around 576 nm at pH 7. This is the food-relevant blue band.
- pH 8 to 11: blue-green to green from deprotonated quinoidal forms.
- pH 12 to 14: yellow-brown, after acyl-bond hydrolysis opens the structure into chalcones.
Citric acid is roughly pKa 3.13. A small amount in a candy melt or beverage drops the pH into the 3 to 4 band. That is purple territory. Lemon juice (pH 2 to 2.5) drops it further into red. The pigment did not break. The equilibrium shifted. The hue followed.
This is also why the change is reversible. Buffer the system back toward pH 6 and the blue comes back. The molecule is the same throughout.
Whereas an Anthocyanin Like Cyanidin-3-Glucoside Would Behave Very Differently
It is worth pulling apart "anthocyanin" as a category. Ternatins are unusual. The more common anthocyanins used in food (cyanidin-3-glucoside from black carrot, grape skin, elderberry) are not polyacylated, do not have the sandwich-stacking stability boost, and follow a different rule set:
- Non-acylated anthocyanins stay red across most of the food-relevant pH range. They only turn blue at pH above 7, where they are highly unstable.
- Stability gap is large. Cabrita et al. measured that the acylated analog petanin retained 84 percent of its color at pH 4 / 10 degrees C after 60 days. Cyanidin-3-glucoside under the same conditions was fully degraded.
- Half-life of pH 3 extracts at 98 degrees C: red sweet potato 4.6 hours, purple carrot 4.6 hours, red grape 2.4 hours, purple corn 2.0 hours. Ternatins under heated conditions outlast all of these because the acyl groups slow hydrolysis.
Practical takeaway: a non-acylated anthocyanin cannot do what butterfly pea does. It cannot deliver a stable blue at food-relevant pH, and it cannot deliver a clean acid-triggered red-to-purple-to-blue gradient. The clean blue is structurally locked to the ternatin chemistry.
Unlike Spirulina, Which Fails for a Completely Different Reason
Spirulina (phycocyanin) is the other major natural blue. Comparing it side by side with butterfly pea is instructive because the failure modes are not the same problem.
- Phycocyanin is a protein-bound chromophore. The blue color comes from phycocyanobilin (a linear tetrapyrrole) covalently attached to an apoprotein. When that protein denatures (typically beginning around 45 to 60 degrees C in aqueous solution and completing above 70 to 80 degrees C), the chromophore loses its color-defining environment. The blue is gone. The protein cannot refold. The change is irreversible.
- Ternatins are a small-molecule equilibrium system. Acid does not destroy them; it shifts the equilibrium toward the protonated forms. Buffer the system back and the blue returns. The change is reversible, at least until heat or alkaline pH starts hydrolyzing the acyl groups and the glycosidic bonds.
Why this matters at the bench: a phycocyanin product that flashes through a 75 degree pasteurization may come out green for good. A butterfly pea product that flashes through a low pH may come out purple now and snap back to blue later if the formulation buffers up. They are different problems and they need different formulation moves.
The Formulation Move: Late Addition, and Why the Numbers Justify It
The standing advice -- add butterfly pea at the very end of the heating cycle -- is real, and the kinetics back it. Jiang et al. 2021 measured ternatin half-life at pH 7 across temperatures:
- 4 degrees C: 334.2 days
- 25 degrees C: 47.1 days
- 50 degrees C: 18.5 days
That is a 2.5x reduction in half-life going from room temperature to 50 degrees, in a still solution. Real heating cycles have oxygen exposure, mechanical agitation, and pH drift, all of which accelerate degradation further. Marpaung et al. 2017 confirmed that ternatin degradation at pH 7 is statistically negligible at 7 degrees C, becomes meaningful at 30 degrees C, and runs hard at 60 to 90 degrees C.
Translation for a formulator: every minute the ternatin spends above 30 degrees C, you are losing color. Add the pigment as close to the end of the heating step as the process permits. If pasteurization is a hard requirement, accept the color hit upfront in your dosing math rather than dosing for fresh and watching the product fade on shelf.
One thing the literature does not support: the claim that late addition lets you use lower ppm. What it does support is better color retention per gram of extract, which often does mean lower effective dosage at finished-product spec, but the mechanism is preservation, not synergy.
Lean Into the Purple, On Purpose
Not every project needs the blue. Some of the best butterfly pea applications use the pH transition as a feature, not a bug. A sour candy that ships pink and turns purple as you suck on it. A cocktail mixer that starts blue in the bottle and turns rose when it hits lemon juice. A sports drink that visibly shifts color across the pH gradient of a fruit garnish.
That dynamic, reversible, acid-triggered color change is the one thing synthetic FD&C dyes cannot replicate in a single ingredient. The static purple endpoint, on its own, can be approximated reasonably well with a blend of FD&C Blue 1 and FD&C Red 40 to a target CIELAB coordinate. The motion cannot. If your product is built around the motion, ternatins are the only practical answer.
If your product needs a stable blue and the pH cannot stay above 5, butterfly pea is the wrong tool. Look at spirulina (with a thermal caveat), or at lipid-soluble pigment systems for fat-continuous matrices.
Where Butterfly Pea Is Now FDA-Approved -- Including the May 2025 Update Most Articles Miss
Butterfly pea flower extract is listed at 21 CFR 73.69 as a color additive exempt from certification. Two events matter:
- September 2, 2021 (86 FR 49230): Original listing. Approved uses included alcoholic beverages, sport and energy drinks, flavored and carbonated water, fruit drinks and smoothies, carbonated soft drinks, fruit juice teas, coated nuts, liquid coffee creamers, ice cream and frozen dairy desserts, hard candy, dairy and non-dairy drinks, fruit preparations in yogurts, and soft candy.
- May 12, 2025 (90 FR 20101), effective June 26, 2025: Expansion adding ready-to-eat cereals, crackers, snack mixes, hard pretzels, plain potato chips (restructured or baked), and plain corn chips, tortilla chips, and multigrain chips. This expansion opens butterfly pea to the dry snack category for the first time.
Use level under both rulings is "consistent with GMP" -- no numeric maximum. Batch certification is not required.
If you are working in a 21 CFR category that is not on either of those lists -- baked goods, meat products, surface decorations beyond the listed categories -- butterfly pea is not currently a compliant choice in the U.S. EU and Codex frameworks have their own approvals to check separately.
What This Means for R&D and Regulatory Teams
Reformulating with butterfly pea is not "swap in for Blue 1." It is a different chemistry that wants a different formulation approach. The decisions worth making before you spin a batch:
- Confirm your finished-product pH lands in the 5 to 7 band, or design around the purple endpoint on purpose.
- Move ternatin addition as late in the heating sequence as the process permits, and dose for the heat exposure you actually have.
- Verify your application category is covered under 21 CFR 73.69 (and the May 2025 expansion if you are in dry snacks).
- Decide whether you actually need the dynamic color change. If yes, ternatins are the answer and nothing else will do it. If no, evaluate phycocyanin or a lipid-soluble system against your thermal and matrix constraints.
How DyeConverter Handles This
DyeConverter maps these decisions for you. You input the application (hard candy, sports drink, frozen dairy, cereal coating), the finished-product pH, the thermal profile, and the shelf-life requirement. The platform returns the natural color alternatives that actually fit those constraints, with the dosage ranges, the regulatory status by region, and the supplier SKUs that have published spec sheets for that application.
For butterfly pea specifically, that means filtering for applications inside the current 21 CFR 73.69 scope, scoring the pH window against your formulation, and ranking ternatin-based suppliers (Sensient, GNT, Givaudan, Oterra) on their published thermal-stability data.
Bench-validate the recommendation. The platform's job is to get you to the right starting point so the bench work is iteration, not search.
Frequently Asked Questions About Butterfly Pea Flower Color
What is the actual pH range where butterfly pea flower stays blue?
Roughly pH 5 to 7. From pH 3 to 5 the extract reads as purple. Below pH 3 it goes red. The "pH 3.8 to 7 is blue" range repeated across food-industry blogs is not what the underlying spectroscopy shows.
Is the pigment in butterfly pea flower an anthocyanin?
Yes, but more precisely it is a series of 15 polyacylated delphinidin glucosides called ternatins (Terahara 1990-1998). The acyl groups stack with the chromophore and stabilize the blue form, which is why ternatins are far more shelf-stable than common anthocyanins like cyanidin-3-glucoside.
Why does my butterfly pea blue turn purple the moment I add citric acid?
The flavylium and quinoidal forms that define anthocyanin color sit in a pH-driven equilibrium. Citric acid drops the pH into the 3 to 5 band where the equilibrium shifts toward the protonated forms that read as purple. It is the same molecule, in a different state, and the change is reversible until heat or alkaline pH starts hydrolyzing the acyl groups.
Does heat hurt butterfly pea color?
Yes. Ternatin half-life at pH 7 collapses from 47 days at 25 degrees Celsius to 18.5 days at 50 degrees Celsius (Jiang et al. 2021). Degradation accelerates measurably above 30 degrees. Late addition in the heating cycle protects the color.
Is butterfly pea flower extract FDA-approved?
Yes. Listed at 21 CFR 73.69 as a color additive exempt from certification. Original approval September 2 2021 covered beverages, candies, ice cream, and yogurts. Use was expanded May 12 2025 (effective June 26 2025) to include ready-to-eat cereals, crackers, snack mixes, hard pretzels, plain potato chips, and plain tortilla, corn, and multigrain chips. Use level is "consistent with GMP" -- no numeric maximum.
Final Thought
Natural color failures are rarely ingredient failures. They are matrix failures. Butterfly pea is one of the most thoroughly characterized natural blues we have. The literature is good, the FDA framework is current, and the formulation moves that protect the color are knowable.
The reason butterfly pea projects fail is that teams treat it like Blue 1 and then are surprised when the chemistry tells them otherwise. Spend the time upfront to understand the molecule, and the bench iteration shrinks from months to weeks.