Why Tea Drinks Turn Brown Without Turning Hazy: Alkalinity, Minerals and Holding Time

Record darkening and haze separately

The same powder lot and the same feed water start producing a darker liquor on a given day, while turbidity readings do not move with it. Filing that batch under “cold haze” or “poor solubility” sends the next investigation away from the axis that actually moved, so alkalinity, mineral-salt composition and holding time stay outside the comparison design. Colour and turbidity track different changes: browning follows soluble colour products and catechin conversion [1][2], haze follows particles, complexation and precipitation [5]. The first step in an investigation is to give every sample a colour reading and a turbidity reading on the same day, then see which axis moved.

Alkalinity decides whether browning occurs

A 2021 teabag-infusion study split its waters into five hardness grades of synthetic freshwater, one sodium bicarbonate solution and one solution containing only the remaining mineral salts, so that alkalinity and salt effects could be separated [1]. All three slightly alkaline liquors browned (sodium bicarbonate at pH 7.36, very hard water at 7.24, argon-sparged very hard water at 7.88), while the deionized control and the mineral-salt control (pH 5.68 and 5.51) showed no browning [1]. In the same experiment EGC and EGCG declined steadily in the alkaline liquors while EC and ECG held up [1], and a separate 1997 model study reports the same stability ordering [6]; those two catechins carry three hydroxyl groups on the B-ring, radicals form preferentially on that structure, and polymerization then yields brown products [1][2]. Bicarbonate in hard water keeps the liquor in the alkaline range after brewing [1].

Mineral salts decide how fast it develops

In the same experiment, the mineral-salt control stayed acidic and did not brown on its own; once the system is alkaline, salts turn into a rate term. The sodium bicarbonate liquor had the highest absorbance at 470 nm during the first two hours, then the very hard water sample (calcium sulfate, magnesium sulfate, potassium chloride) overtook it, with faster EGC and EGCG decay as well [1]. Calcium and magnesium also run a separate path: a 2013 experiment reported that green tea infusion became more turbid above 40 mg/L Ca2+ in the brewing water, with turbidity negatively correlated with oxalic, quinic and tartaric acid contents [5] (that experiment’s conditions). A water change therefore has to be read on two axes, and a single hardness or TDS value cannot say where the colour problem came from.

Holding time decides how much you see

Browning accumulates with time. In the 2021 study the sodium bicarbonate liquor led the 470 nm absorbance for the first two hours before the very hard water sample passed it; in that very hard water tea, about 3 mg of EGC and 2 mg of EGCG were lost per hour [1] (figures for that system). On the finished-drink side, a 2026 storage study of green tea beverages sampled at 4 °C, room temperature, 37 °C and 55 °C over 3 to 56 days found total colour difference rising with temperature and time, L* falling continuously, total polyphenols declining markedly and caffeine and free amino acids changing insignificantly [3]; those same drinks stayed clear, with no notable tea cream or sediment [3]. Browning can therefore occur on its own while the drink stays clear, and recording how long the liquor stood before filling, and at what temperature, is what separates accumulated time from raw-material variation.

More than one route to a brown liquor

Colour darkening has at least two sources. The storage study listed oxidation and degradation of phenolic compounds alongside coloured Maillard reaction products as the primary drivers of colour change, reported a marked decrease in catechins, especially EGCG, in support of phenolic oxidation, and noted that free amino acids did not change significantly under any storage condition, which points away from Maillard as the main contributor in that system [3]. Model-system work shows that, without enzymes, auto-oxidation of EC and EGC produces dimers and oligomers with an extended conjugated system, fused rings and carbonyl groups, and those are responsible for the brown colour [2]; EGCG auto-oxidation and epimerization respond to temperature, pH, oxygen partial pressure and antioxidant level [9]. The control points therefore fall into two groups: temperature and time on one side, pH and oxygen exposure on the other.

Read all three the same day: colour alone will mislead

Three things happen for each sample on the same day: absorbance is read over the 400–500 nm range at a fixed path length and a fixed measurement temperature, which is where brown compounds absorb [1][2]; turbidity is read by one consistent method; and EGC, EGCG, EC, ECG and caffeine are checked by HPLC or by an outsourced profile [1]. The reading rules go on the record sheet in advance: colour up and turbidity flat sends the investigation to final pH, feed-water alkalinity and holding time; turbidity up and colour flat sends it to calcium and magnesium, oxalate-type anions and cooling conditions [5]; both up means fixing pH by acidification first and then comparing salt composition, changing one variable per trial.

Keep two record layers: feed water and tea liquor

Water records and liquor records are filled in separately. The water line carries calcium, magnesium, total hardness with units, alkalinity with its reporting basis, pH, conductivity, sampling point and sampling time; the liquor line carries pH after reconstitution, after recipe acidification and after heat treatment, plus temperature, sampling time, colour and turbidity [1][3]. Raw-water pH cannot stand in for liquor pH: in the same study the synthetic waters ran from pH 8.3 down to 7.2 as bicarbonate fell, brewed tea liquors sat between pH 7.1 and 6.0, and brewing itself lowers pH further as amino acids and other organic acids come out of the leaves [1]. Whether a sample is alkaline is judged from the liquor reading.

A holding-time sampling template (hypothetical example)

The time points and criteria below are hypothetical numbers for illustration only — not batch data: draw samples at 0, 1, 2, 4 and 8 hours from one prepared batch, bring every sample back to one measurement temperature before reading, record colour as ΔE or as absorbance at a fixed wavelength, record turbidity alongside, and keep one sample for the catechin profile. The 2021 study obtained its EGC and EGCG decay curves precisely by re-injecting the same liquors repeatedly over four hours [1]. Both ends of the sampling window come from the plant: how long the liquor stands before filling and how long it circulates in process, both belong inside the window, with oxygen exposure, sampling temperature and container identity recorded next to them. Fixed temperature and fixed wavelength are what make “darker” comparable, and what keeps a concentration difference from being read as browning.

Qualify extraction and stability separately

“Reconstitution yield” and “what is left after standing” are two different acceptance items. A brewing-water study reported that high-conductivity water reduced extraction of catechins and caffeine while high pH mainly affected catechin stability [7]; the 2021 water-hardness study likewise observed that alkaline conditions generally raised extraction efficiency, with caffeine and the most stable catechin ECG yielding more in the sodium bicarbonate liquor while EGC and EGCG decayed fastest [1]. Testing only the initial content misses the holding period, and testing only the end point mistakes poor extraction for degradation. When talking to an ingredient supplier, ask three things separately: the catechin and caffeine baseline of the same powder lot, the reconstitution procedure, and what remains after several hours of standing.

Water studies disagree, so the comparison crosses raw material with water

Water studies do not point in one direction. The 2021 study records in its discussion that Zhou and coworkers reported higher EGCG and ECG contents for green tea prepared with alkaline tap and carbon-adsorbed water than with more acidic water of low ionic strength [1]; the 2009 abstract of that report states that carbon-adsorbed water raised ester-catechin concentrations while deionized water gave the highest yield [8]. Water composition, tea raw material and measurement timing all shift the outcome, so a crossed raw-material-by-water comparison stays in the design. Extrapolation needs boundaries as well: the 2021 study’s “browning commences above pH 6.3 and 42 ppm CaCO3” belongs to a teabag system brewed at 70 °C for 3 minutes [1], the near-complete degradation within minutes above pH 8 comes from a different model solution [6], and the greater stability around pH 4 is an abstract-level result from a concentrate system [4]. Those numbers design the comparison; acceptance limits for a finished drink still have to come from the actual recipe and the actual water.

Next: complete the three records, then run the comparison

When a “colour got darker” report arrives, three things come first: a full analysis of the same water lot, with alkalinity and its reporting basis listed on their own line; the carrier and tea-solids statement for the same powder lot; and a three-axis record of colour, turbidity and catechins taken right after preparation and again after several hours at one temperature. With those in hand the question usually narrows to four directions — change the water or acidify, shorten holding time, change the reconstitution route, change the powder — and only then does it go into a beverage application development discussion with the comparison attached. No direct study covers this three-factor combination of reconstitution, alkalinity or mineral salts, and holding time; run it as an engineering proposal to be validated and keep the records. For cold haze and sediment, see “Why Tea Drinks Turn Cloudy in the Fridge”; for material form and specifications, the green tea ingredient page; for RTD purchasing, the buying checklist.