Ube Powder for Beverage Formulation: Color, Flavor, Dispersibility and Stability

01 Framing

Two Systems, Not One

Nearly every complaint about ube in drinks falls into one of the rows below. The useful thing about the split is that starch problems are physics and colour problems are chemistry. Adding mesh does nothing for a pH problem. Lowering pH does nothing for retrogradation. Teams lose weeks because they treat "my ube drink looks wrong" as one issue.

02 Starch

Why Ube Settles, Thickens and Turns Gritty

Ube powder is mostly starch, and the fraction that matters is amylose, because amylose is what recrystallises.

Why the published amylose numbers do not agree

You will find Dioscorea alata amylose reported at 25.77 percent,^[2] at 34.69 percent,^[3] at 36.2 percent after defatting against 24.2 percent for cassava,^[4] and at up to 40.24 percent in one of five Chinese accessions,^[19] while whole tubers have been reported at 25 to 30 g per 100 g dry basis.^[1] That is a wide spread for one species, and the spread itself is informative.

Three things drive it. Isolated starch and whole tuber flour are different bases, and flour is diluted by protein, fibre and ash. Defatting matters because amylose-lipid complexes do not bind iodine, so an undefatted sample under-reports. And D. alata carries a high proportion of long amylopectin chains^[2] which also bind iodine, inflating apparent amylose in the standard colorimetric assay.

The practical point for a buyer: an amylose figure on a certificate of analysis is only comparable against another figure produced the same way, on the same basis. Ask which.

Either way this is a high-amylose starch by tuber standards, and everything else in this section follows from that.

Gelatinisation temperature decides whether cold water does anything

Native D. alata starch has a gelatinisation onset at 69.62 °C and a peak at 74.81 °C, with a pasting temperature of 72.47 °C.^[2] Below roughly 70 °C the granules do not open. They wet at the surface, swell slightly, and stay granules.

That is why a raw, air-dried ube powder behaves so badly in an iced drink. It is not that the mesh is too coarse. The starch never gelatinised, so it cannot build a continuous phase, and it sits in the glass as suspended solids waiting for gravity.

Drum drying changes this. Drum-dried purple tuber flours are pre-gelatinised, confirmed by both RVA and DSC, with the granule structure disrupted into aggregates and crystallinity no longer detectable.^[5][6] A pre-gelatinised powder hydrates in cold liquid. It also thickens in cold liquid, and it brings a second problem with it.

The part nobody tells you: pre-gelatinisation makes settling worse

In the same study that measured the pre-gelatinisation, drum drying produced aggregated granules of 28.94 to 52.90 µm, while sun drying preserved intact granules of 12.50 to 23.16 µm.^[5]

Settling velocity under Stokes' law scales with the square of particle diameter. Take the midpoints of those two ranges, about 45 µm against 18 µm, and the larger particles fall roughly six times faster.

Read that last line again, because it is the whole argument. Thickening a beverage fivefold, enough that a consumer notices, buys you hours. Your shelf life is months.

A fluid with a genuine yield stress holds a particle indefinitely as long as the gravitational stress on it stays below that threshold. This is why low-acyl gellan gum at 0.02 to 0.05 percent works in a water-thin beverage: it forms a weak, shear-reversible gel network with a measurable yield stress while barely changing perceived thickness. It is also why xanthan at comparable viscosity disappoints. Xanthan is strongly shear-thinning but has a much weaker yield stress, so it slows settling without stopping it.

Which leads to the uncomfortable conclusion for a drum-dried powder, ours included. Pre-gelatinisation solves hydration and makes suspension harder. A pre-gelatinised ube powder in a low-viscosity cold-fill RTD needs a suspending system more than a raw powder would, not less. Anyone claiming their ube powder stays suspended on its own in a water-thin beverage has either already blended in a hydrocolloid or has not run a six-month trial.

The assumptions in that calculation are soft. Particle density in a hydrated aggregate is hard to pin down, real suspensions flocculate, and hindered settling changes the arithmetic at high solids. Treat the numbers as order of magnitude. The ranking between the two particle sizes is robust; the absolute minutes are not.

Retrogradation, and why 4 °C is the worst place to store the product

D. alata starch shows a setback viscosity of 755.75 cP against a peak of 2031 cP,^[2] and the high-amylose varieties have been described as giving greater gel firmness and stronger retrogradation potential.^[3] Setback is viscosity regained on cooling and is a direct proxy for how readily the starch recrystallises.

Two timescales are at work. Amylose retrogrades fast, over hours, and is essentially irreversible at beverage temperatures. Amylopectin retrogrades slowly, over days to weeks, and can be partly reversed by reheating. A high-amylose starch means the fast, irreversible one dominates.

The temperature dependence is the part that catches formulation teams out. Retrogradation rate peaks not at ambient and not at freezing but in the chill range, because nucleation and chain mobility are both adequate there. A refrigerated RTD sits very close to the worst possible temperature for this reaction across its entire shelf life.

Three practical readings follow. A frozen smoothie base or an ambient shelf-stable format is kinder to ube than a chilled one. A chilled RTD needs its stability trial run at 4 °C rather than at elevated temperature, because the usual 37 °C accelerated protocol will understate the grit. And a hot latte that tastes perfect at the counter tells you nothing about the same formula held twelve hours cold.

Resistant starch, which is a label question and a dose question

Native D. alata starches carry high resistant starch, higher than D. opposita, D. esculenta and D. nipponica. In a comparison of five Chinese D. alata accessions, the highest-amylose accession reported RS2 at 84.17 percent and RS3 at 10.48 percent.^[19]

Gelatinisation destroys RS2, since RS2 is the native granular structure. So a drum-dried powder starts with far less resistant starch than the raw tuber. Then retrogradation regenerates RS3 during chilled storage, which means the same reaction that creates the grit creates the resistant starch.

Two consequences worth knowing. If you intend a dietary fibre claim based on resistant starch, the number will drift over shelf life and is not something to declare from a raw-material figure. And at high inclusion in a beverage, unexpected fermentable substrate reaching the colon can produce consumer complaints that get misattributed to lactose or to the sweetener.

The gluey ceiling

Pre-gelatinised ube starch builds viscosity immediately on hydration. In a 240 mL latte most formulators land between 8 and 15 g of powder, depending on whether they want colour plus body plus some flavour, or colour alone. Above roughly 6 to 7 percent w/v the drink starts reading as thin porridge, and the starch also begins to mute aroma release, which is why heavily dosed ube drinks taste flatter than the powder smells.

Those are common industry starting points, not validated figures for your system. Milk solids, sweetener load, homogenisation pressure and fill temperature all move the ceiling.

03 Colour

Ube Is More Stable at Neutral pH Than the Textbook Says, With a Caveat

This is where most published advice about ube goes wrong, because it applies generic anthocyanin behaviour to a pigment that is not generic. It is also where supplier marketing overstates the case in the other direction, so both corrections are below.

The generic rule

Anthocyanins sit in a pH-dependent equilibrium. Below about pH 3 the flavylium cation dominates and gives a clean red-purple. Between roughly pH 4 and 6 the colourless carbinol pseudobase takes over and intensity drops sharply. Near neutral the quinoidal base gives blue-purple. Higher still and you get chalcone and yellow-brown breakdown. In purple sweet potato extract, pH was found to matter more for stability than temperature did.^[7]

By that rule, dairy at pH 6.7 should be hostile to any anthocyanin.

Why ube is partly an exception

The anthocyanins of D. alata are acylated. Six have been isolated from one characterised accession, all cyanidin derivatives, with alatanin C, cyanidin 3-(6-sinapoyl gentiobioside), accounting for around 70.5 percent of the total.^[8] Alatanin C was identified as the first monoacylated anthocyanin found to be unusually stable in neutral aqueous solution. The stability was attributed to two distinct effects: intramolecular stacking of the sinapic acid moiety over the chromophore, and chiral self-association of the anthocyanidin nuclei.^[9]

Miyata and colleagues tested purple yam against reference materials and found higher thermostability than both purple sweet potato and purple potato, with alatanin A identified by quantitative HPLC as a major contributor.^[10]

The caveat that nobody publishes

Two things need saying about how far that evidence stretches.

First, Miyata measured thermostability, at a given condition. Thermal stability and pH stability are not the same property, and a paper demonstrating one does not demonstrate the other. Suppliers routinely cite this study as proof that ube holds colour in milk. It is not that.

Second, and more useful to you: of the two stabilising mechanisms in the Yoshida work, only one survives dilution. Intramolecular stacking is a property of the single molecule and holds at any concentration. Self-association is intermolecular and therefore concentration-dependent. In a purified model solution at the concentrations used for structural work, self-association contributes substantially. In a finished beverage, where the pigment is orders of magnitude more dilute, that contribution largely disappears.

So the honest formulation statement. Ube outperforms butterfly pea, elderberry, black carrot and most fruit anthocyanins at neutral pH, because those are largely unacylated. It will still shift toward blue-violet at pH 6.7 and it will still lose intensity there. It just will not collapse.

If your ube latte has gone properly grey rather than blue-violet, pH is probably not your culprit. Look at section 04.

Copigmentation, the lever most ube formulators never touch

Copigmentation is the strongest tool available for anthocyanin colour in beverages and it goes almost entirely unmentioned in ube application literature. A colourless phenolic stacks against the flavylium chromophore, shields it from nucleophilic water attack, and produces both a bathochromic shift toward blue-purple and a hyperchromic increase in intensity. It is the same physics as the intramolecular stacking above, borrowed from a neighbour molecule instead of a covalently attached one.

Ube brings its own copigment pool. Thai D. alata tubers carried total phenolics of 100 to 385 mg gallic acid equivalent and flavonoids of 60 to 160 mg catechin equivalent per 100 g dry basis.^[1] A whole tuber powder delivers those alongside the pigment. An anthocyanin extract does not, which is one reason a purified ube colour can look weaker in application than the powder it came from.

Three practical levers. A small addition of a polyphenol-rich ingredient such as green tea extract, rosemary extract or grape skin will deepen and blue-shift the colour at doses well below their own flavour threshold. Sugar at beverage concentrations assists modestly by lowering water activity. And processing that strips phenolics, meaning heavy filtration, adsorbent resins or aggressive clarification, costs you colour that the anthocyanin assay will not show as lost.

We are not aware of a published copigmentation study run specifically on D. alata anthocyanins in a dairy matrix. The mechanism is general and well established; the magnitude for this pigment in your system is a bench question.

Specifying colour as a number

"Purple" is not a specification. If colour consistency matters, measure in CIELAB and write a tolerance.

Ube violet sits in the fourth quadrant of the a*b* plane, positive a* and negative b*, so hue angle falls between 270° and 360°. A blue shift in dairy moves hue angle down toward 290 to 300°. Browning from enzymatic activity or heat pulls b* positive and swings the hue angle out of the quadrant entirely, which is why a browned lot reads as muddy rather than pale.

Set your own targets rather than borrowing ours or anyone else's, because the value depends on your matrix, your fat level and your instrument geometry. The workable approach is to measure a reference batch you are happy with, fix L*, a* and b* as the target, and specify incoming lots and finished product against a ΔE tolerance. Chroma tells you intensity, hue angle tells you whether the shade drifted. Most teams track only one of the two and then cannot explain what changed. How to judge incoming lots wet rather than dry, and why dry powder always looks paler than the drink it makes, is covered in ube powder not purple?

Flavour, briefly

Ube's aroma is mild, vanilla with a nutty note, sometimes read as faintly coconut. It is not high-impact and it does not survive dilution the way a flavouring compound does. Formulators wanting a recognisable ube taste at a colour-appropriate dose generally pair the powder with a small flavouring addition and accept that the powder is doing colour and body. That is a legitimate formulation choice. What you call it on the label is a separate question, covered in our comparison of ube powder, ube extract and ube flavouring.

04 Colour killers

Four Things in Your Formula, and One in Your Supplier's Factory

Ascorbic acid, irreversible and the most common self-inflicted wound

Ascorbic acid degrades anthocyanins by two proposed routes: direct condensation at the C-4 position of the anthocyanin, and autoxidation of the ascorbate to generate hydrogen peroxide, which opens the pyrylium ring to a chalcone.^[11][12] Both molecules are consumed. This is described as reciprocal and irreversible destruction of both the pigment and the micronutrient, and it is explicitly distinguished from sulfite bleaching, which is reversible.^[13]

If your brief reads purple, natural, and a vitamin C claim, something has to give. The options are encapsulating the ascorbate, switching to an ascorbic acid derivative, since 2-O-phosphate and 2-O-sulfate esters have been shown to bleach anthocyanins far more slowly than free ascorbic acid,^[14] or moving the vitamin C claim to a different SKU. Adding more anthocyanin to compensate does not work. You are feeding the reaction.

Sulfites, and what their presence tells you about the supplier

Bisulfite attacks the flavylium cation to form a colourless adduct. Jurd established in 1964 that this is reversible, does not hydrolyse the glycoside and does not reduce the pigment; the bisulfite ion simply adds to the carbonium ion, giving a colourless product structurally similar to the carbinol base.^[15]

The immediate consequence is that a sulfited powder can assay acceptably for total anthocyanin while looking visibly pale, because a portion of the pigment is locked in a colourless form. Buyers read this as a weak lot. It is not weak. It is bleached.

The more interesting question is why sulfites are there at all, and the answer points at thermal processing.

Yam is unusually rich in polyphenol oxidase, and D. alata is worse than its relatives. PPO activity was 50 percent higher in non-processed freeze-dried D. alata than in D. rotundata.^[16] PPO oxidises the tuber's own phenolics to quinones, which polymerise to brown pigments, and those browns sit on top of the purple and read as a grey or muddy cast.

Blanching is the standard control, but gentle blanching does not work. In the same study, 45 percent of PPO activity survived 50 minutes of blanching at 60 or 65 °C.^[16] Peroxidase dropped below 20 percent within 10 minutes at any temperature tested, but peroxidase has a second trick. Only POD regenerates activity after thermoinactivation, because its inactivation is conformational while PPO's involves covalent damage, and POD reactivation is thought to contribute materially to browning in processed yam.^[17]

The questions to put to any supplier: what is the temperature and hold time of the cook or blanch step, is peroxidase activity checked as the endpoint, since a POD-negative result is the standard industry indicator that blanching was adequate and it is a cheap test, and what is the residual SO₂ figure on the certificate of analysis rather than just a sulfite-free claim in the marketing copy.

For transparency about our own material: our powder is made from Dioscorea alata sourced from the Philippines, from peeled tuber flesh steam-cooked at 105 °C before drum drying, with no sulfites used at any stage. The trade-off is real. That route narrows the processing window, demands tighter control of the interval between peeling and cook, and costs more than dipping the slices in metabisulfite would. We think that is the right trade for a pigment product. Buyers who need the cheapest available purple tuber powder are not our customers.

Metal ions

Cyanidin carries an ortho-dihydroxy B-ring, which chelates iron and aluminium and shifts the colour blue-green. Since alatanin C and the rest of the ube profile are cyanidin derivatives,^[8] ube is fully exposed to this.

Sources in order of how often they turn out to be the problem: a ferrous salt in a mineral or vitamin premix, which is by far the most common and is often overlooked because the premix is dosed at a few hundred ppm and nobody suspects it; process water; and equipment, particularly older or damaged stainless and any aluminium contact.

For reference points on water, drinking water iron is limited to 0.2 mg/L under the EU Drinking Water Directive and 0.3 mg/L as a US secondary standard. Those are aesthetic thresholds for water, not established thresholds for anthocyanin discolouration, and we have not found a published figure for the level at which ube specifically shifts. Treat it as a diagnostic path rather than a number. If the drink reads slate, the pH checks out and there is no ascorbate in the formula, run a water analysis and a premix trial before you blame the powder.

Heat load, not peak temperature

The drying data makes this concrete. Drum drying, which is high temperature and short residence, gave purple sweet potato flour with anthocyanin retention of 1.05 to 7.57 mg/g against 0.59 to 1.17 mg/g for sun drying at ambient temperature over days.^[5] Expressed as loss, drum drying cost 35 percent and 75 percent across two varieties while sun drying cost 90 percent and 86 percent.^[5]

Those figures are for Ipomoea batatas, not Dioscorea alata. We have not found an equivalent published comparison for purple yam. The principle transfers; the numbers should not be quoted as yam data, and we would rather flag that than let a convenient statistic travel further than its evidence.

For your process the same logic says a short HTST pass is kinder than a long hold at 85 °C, and that a deaerated fill buys you more colour than shaving five degrees off pasteurisation.

The one that helps: protein

Milk protein is not purely an enemy. Wheat protein hydrolysates complexed with purple sweet potato anthocyanins increased the thermal degradation half-life 1.65-fold after sterilisation at 121 °C and pH 6.8. After 7 days at 37 °C and 45 °C, retention was 5.4 and 32.2 times higher than the anthocyanins alone, and colour change fell from ΔE 6.19 and 10.46 to 0.29 and 0.77.^[18]

Sweet potato pigment again rather than yam, so direction rather than specification. The practical read is that dispersing ube powder into the protein phase before heat treatment is likely to protect colour better than dosing it into an already-heated base. That costs nothing to test on your own line and it is the single cheapest experiment on this page.

05 Lot variation

Where Batch Colour Difference Comes From, and How to Buy Around It

Buyers usually assume a visible colour difference between lots means the supplier substituted or cut the material. Occasionally true. Much more often the tuber was simply different.

Two Thai D. alata accessions sampled across vine ages of 3 to 8 months gave monomeric anthocyanin of 10 to 90 mg cyanidin-3-glucoside equivalent per 100 g dry basis, depending on accession and age.^[1] That is a ninefold range inside one species. The same work found the type and content of the minor alatanins was accession-dependent and could itself influence the intensity of the purple in the resulting flour.^[1] Work on five Chinese D. alata accessions found the same pattern in the starch fraction, with amylose, resistant starch, gelatinisation temperature and crystal type all varying between accessions of the same species.^[19]

We are deliberately not stating whether the pigment number rises or falls with maturity. The published abstract reports the range and the dependence without giving the direction, and we have not read the full dataset. Anyone telling you confidently that later harvest means deeper colour should be asked for their source.

What is certain is that cultivar, harvest timing, soil and season all move it, and that no amount of process control at the dryer recovers pigment that was never in the tuber.

Dose on colour value, not on weight

This is standard practice with natural colorants and it is the only approach that produces a consistent finished product. Most beverage teams resist it because it complicates batching, then spend a year chasing colour drift instead.

Then notice the problem that arithmetic creates, because this is the part that gets missed. You have just increased the starch load by 31 percent as well. At 15.7 g in 240 mL you are at 6.5 percent w/v and pushing the gluey ceiling from section 02.

So colour-value dosing only works inside a band. Set an incoming specification with a floor, not just a target. A lot assaying below roughly 75 to 80 percent of target cannot be corrected by dosing up without changing the texture of the drink, and should be rejected or blended rather than compensated. Suppliers who will not give you a floor are asking you to absorb their raw material variability in your texture.

What to ask the certificate of analysis to actually say

• Total monomeric anthocyanin by the pH differential method, with the method named. It is cheap and it is the industry default. Note its blind spot: the assay reads at pH 1.0, which drives bisulfite adducts back to the coloured flavylium form, so a sulfited powder can report a respectable number while looking pale in your product. Assay and appearance disagreeing is itself a signal.
• An HPLC profile if colour consistency is critical. It tells you whether the alatanin distribution is stable lot to lot, and it distinguishes a cyanidin-dominant yam profile from the peonidin-heavy profile of purple sweet potato.
• Residual SO₂, as a number. Above 10 mg/kg it is a declarable allergen in both the EU and the US, and the figure also tells you something about the thermal process behind it.
• Particle size distribution, not a single mesh figure. Section 02 explains why the distribution, not the nominal cut, drives your settling behaviour.
• Cultivar and harvest window, if you can get them written into the specification. Single-origin material from one base on one harvest schedule varies less than material aggregated across a region, and that is most of what you are paying for when you pay more.
• A retained sample from a previous lot. A supplier who keeps retains can show you the real range you are buying into instead of a photograph of their best batch.

06 Carrier and wetting

Two Questions That Get Confused With Each Other

Why spray-dried ube always has a carrier

Purple yam is a starchy tuber, not a juice. That sentence explains most of the carrier question.

To spray-dry ube you must first make a pumpable slurry, and to get that slurry through an atomiser and off the chamber wall as free-flowing powder you need a drying aid, normally maltodextrin or a modified starch. The carrier is not an additive bolted on afterwards. It is what makes the process physically possible. A spray-dried ube powder is therefore a blend, and the ube fraction is often 50 to 70 percent. Some retail products declare it plainly, with ube and modified starch as encapsulator appearing on at least one widely sold brand's ingredient statement.

Drum drying and hot-air drying both produce powder directly from tuber flesh and need no carrier at all.

How to tell whether the powder you were sent has one

This is what buyers actually need and it is almost never written down. Maltodextrin is essentially pure carbohydrate, with no protein, negligible ash and no fat. Tuber powder is not.

• Protein and ash arithmetic. Pure purple yam powder carries meaningful protein and ash. A 50 percent maltodextrin blend halves both. Compare the declared figures against a reference for pure material. A suspiciously low ash figure is the easiest tell, because nothing else in the process removes minerals.
• Reducing sugar or DE. Maltodextrin has a measurable dextrose equivalent. Whole tuber powder essentially does not. This is a direct test rather than an inference.
• Anthocyanin per gram. Dilution is linear. If the pigment figure is roughly half what comparable material reports, something is diluting it.
• Ask for the process flow diagram. If the answer is spray drying, there is a carrier, whatever the specification sheet says about purity.

When a carrier earns its place, and when it costs you

It earns it for free-flowing behaviour in automated dosing, caking resistance in a humid warehouse, uniformity in a pre-blend where ube is one of eight ingredients, and genuine encapsulation to protect pigment from oxygen over a long shelf life.

It costs you on clean-label positioning, on dilution of colour value so you dose more powder to reach the same shade, and on an added carbohydrate you did not want on the nutrition panel.

One nuance is worth knowing, because "maltodextrin is hygroscopic" is stated everywhere and is too crude to act on. Hygroscopicity tracks dextrose equivalent. Low-DE maltodextrin, roughly DE 10 to 15, has a higher glass transition temperature and resists moisture pickup and caking well. High-DE material, DE 18 to 20 and above, has a lower Tg, picks up moisture faster and cakes sooner. A carrier chosen for cost rather than DE can make storage stability worse than the uncarried powder it was supposed to protect. If you are buying a carrier blend, the DE belongs in your specification.

Wetting is not solubility

Ube powder does not dissolve. Starch is not soluble in water at any temperature. A pre-gelatinised powder forms a dispersion that behaves like a solution to the eye, which is a different thing.

The lumping problem follows directly from pre-gelatinisation. Dry powder hitting liquid hydrates at the surface instantly, forming a gel skin around a dry core, the classic fish-eye. Once that skin forms, no amount of stirring gets liquid inside.

Three fixes, in order of how well they work. Dry-blend the powder with the sweetener at 3:1 or higher before it touches liquid, since the sugar crystals physically separate the particles and this solves the problem outright in most café and sachet applications. Add the powder into a vortex under shear rather than onto a still surface. Or buy agglomerated material, where controlled granulation creates porous clusters that wick liquid inward instead of skinning.

One correction to advice we have given before and seen elsewhere. Warming the liquid to 40 to 50 °C is not necessary and is mildly counterproductive. A pre-gelatinised powder does not need heat to hydrate, and warmer liquid accelerates surface hydration, which makes skinning slightly worse. Disperse at whatever temperature the process wants, and solve wetting with dry-blending and shear instead.

07 By application

Starting Points for Four Formats

Every figure is a starting point for bench work. If you take one thing from the table, make it the RTD column. Chilled storage is the hardest case for ube on both axes at once, because 4 °C is near the retrogradation maximum and a long shelf life is where pigment losses accumulate.

Working on an ube beverage

Our purple yam (ube) powder is made from Dioscorea alata sourced from the Philippines, drum-dried from peeled tuber flesh with no carrier and no sulfites. The certificate of analysis carries an HPLC anthocyanin result. Samples are free of charge, MOQ is 1 kg, lead time is 7 to 15 days and quotations are returned within 24 hours.

Tell us the format, the pH and the coldest temperature the product will see, and we will tell you which of the failure modes above your formulation is exposed to, including where the answer is that a different ingredient suits you better. For a formulation project, ask for two different lots rather than one, so you can see the variation described in section 05 before you write your specification rather than after.

08 Questions

Frequently Asked Questions

09 Evidence base

References

[1] Srivichai, S., Hongsprabhas, P. (2020). Profiling anthocyanins in Thai purple yams (Dioscorea alata L.). International Journal of Food Science, 2020, 1594291. DOI: 10.1155/2020/1594291

[2] Isolation and characterization of yam (Dioscorea alata L.) starch from Brazil (2021). LWT – Food Science and Technology. View source

[3] Physicochemical, pasting and thermal properties of the starch of three varieties of yam (Dioscorea spp.) as potential food ingredients. Polymers, 18(8), 943. DOI: 10.3390/polym18080943

[4] Amylose contents, rheological properties and gelatinization kinetics of yam (Dioscorea alata) and cassava (Manihot utilissima) starches (2003). Carbohydrate Polymers. View source

[5] Influence of drying approaches on the structural, nutritional, and functional characteristics of purple sweet potato flour. CyTA – Journal of Food, 24(1). DOI: 10.1080/19476337.2026.2672215

[6] Effect of drying conditions on properties, pigments and antioxidant activity retentions of pretreated orange and purple-fleshed sweet potato flours (2015). Journal of Food Science and Technology. DOI: 10.1007/s13197-015-2086-7

[7] Stability and quality of anthocyanin in purple sweet potato extracts (2019). Foods, 8(9), 393. DOI: 10.3390/foods8090393

[8] Identification of anthocyanins in a purple yam (Dioscorea alata) accession and their in vitro antiproliferative activity (2023). Journal of Plant Biochemistry and Biotechnology. DOI: 10.1007/s13562-023-00828-2

[9] Yoshida, K., Kondo, T., Goto, T. (1991). Unusually stable monoacylated anthocyanin from purple yam Dioscorea alata. Tetrahedron Letters, 32, 5575–5578. View source

[10] Miyata, R., Sano, H., Hoshino, S., Kumazawa, S. (2022). Thermostability and catechol-O-methyltransferase inhibitory activity of acylated anthocyanins from purple yam. Bioscience, Biotechnology, and Biochemistry, 86(7), 916–921. DOI: 10.1093/bbb/zbac060

[11] Factors affecting the stability of anthocyanins and strategies for improving their stability: a review (2024). Food Chemistry: X. View source

[12] The influence of chemical structure and the presence of ascorbic acid on anthocyanins stability and spectral properties in purified model systems (2019). Foods, 8(6), 207. DOI: 10.3390/foods8060207

[13] Investigating the interaction of ascorbic acid with anthocyanins and pyranoanthocyanins (2018). Molecules, 23(3), 744. View source

[14] US Patent 4,208,434. Color stable food containing anthocyanic pigments and bio-available vitamin C. View source

[15] Jurd, L. (1964). Reactions involved in sulfite bleaching of anthocyanins. Journal of Food Science, 29, 16–19. DOI: 10.1111/j.1365-2621.1964.tb01685.x

[16] Biochemical origin of browning during the processing of fresh yam (Dioscorea spp.) into dried product (2005). Journal of Agricultural and Food Chemistry. View source

[17] Chilaka, F. C., et al. (2002). Browning in processed yams: peroxidase or polyphenol oxidase? Journal of the Science of Food and Agriculture. DOI: 10.1002/jsfa.1119

[18] Feng, Y., et al. (2024). Wheat protein hydrolysates improving the stability of purple sweet potato anthocyanins under neutral pH after commercial sterilization at 121 °C. Foods, 13(6), 843. DOI: 10.3390/foods13060843

[19] Physicochemical characterizations of five Dioscorea alata L. starches from China (2023). International Journal of Biological Macromolecules. View source