Surface Film or Bottom Sediment? Two Mechanisms, Two Water Checks for Tea Scum

A complaint comes in: a brittle film around the rim while the liquor underneath stays clear. Another bottle from the same batch shows a sediment at the bottom and no film anywhere. Both get filed as one solubility problem, and the response — swap the powder, acidify harder, filter longer — fixes neither. The film grows at the air–liquid surface; the sediment forms in the bulk. Each follows its own chemistry, so the first task is attribution: which fault am I looking at, and does the water explain it?

The starting classification already exists: Why Tea Drinks Turn Cloudy in the Fridge separates cold haze, powder lumps and surface film and runs the eight-condition water × reconstitution × storage screen. This article takes the surface-film branch one step further — how to pin a rim film to the water inside one working day, which water numbers matter for it, and what to change first once the fault is assigned.

Where the fault sits: surface versus bulk

Scroll tables horizontally; focus a table region and use the arrow keys.

Question	Rim film (tea scum)	Bottom sediment / cold haze (tea cream → sediment)
Where it grows	Air–liquid interface, on the surface and the vessel rim	Bulk liquid, appearing on cooling or concentration
What drives it	Holding temperature, oxygen, exposed surface area [S06]	Cooling, dilution, cold storage
Required ions	Ca²⁺ (or Mg²⁺) and HCO₃⁻, at a pH high enough to keep the reaction going [S01][S06]	Ca²⁺ alongside polyphenols, caffeine and oxalate-type anions
What it is made of	High-molecular-weight organic material with islands of calcium carbonate or bicarbonates of magnesium or manganese [S01]	Pigmented phenols: thearubigins ≈86%, theaflavins ≈12%, flavonol glycosides ≈2% [S10]
Fastest identification	A drop of dilute acid on a film scraped from the rim: effervescence means carbonate [S05]	Gentle warming: falling turbidity means reversible material [S14]
Effect of filtering the liquid	Ordinary clarification removes existing bulk particles; prevention of later film is not demonstrated here when dissolved precursors and holding conditions remain unchanged. Interface location alone does not settle this [S06]	Only postponed: see the filtration rows below [S09][S18]

Two asymmetries are worth carrying into the lab. Caffeine matters for the two faults in opposite directions: decaffeinated tea forms about as much film as regular black tea [S06], while in the bulk theaflavin seeds clusters of roughly 3 nm and caffeine fills the gaps [S04] — and in a dialysed green-tea infusion, EGCG or caffeine alone stayed clear whereas the two together produced visible haze [S02]. Metal ions behave the same way: calcium drives both, but it arrives with different partners — bicarbonate at the surface, oxalate and polyphenols in the liquid.

What the water has to provide for a film

Three conditions have to be present together: calcium (or magnesium), bicarbonate, and a pH high enough for the reaction to run. Tea brewed with pure Milli-Q water showed no tea film formation, while the strongest tea film formed on tea brewed with hard water containing 200 mg CaCO₃ per L [S01]. The same paper describes the film as high-molecular-weight organic material carrying carbonate islands, which is why an acid drop on a film scraped from the rim is a usable field test. A parallel series of spiking experiments in distilled water confirms the pairing: film appears only when both calcium and bicarbonate are present, and water carrying calcium with sulphate instead of bicarbonate — permanent hardness — grows none [S06][S07] (abstract-level evidence from a 1994–1996 series).

That pairing gives a specific reason to keep alkalinity on its own line next to total hardness. Total hardness tells you how much calcium and magnesium are in the water. Alkalinity is aggregate acid-neutralising capacity, not a direct bicarbonate assay or a synonym for temporary hardness. In bicarbonate-dominated water of known composition and pH it can inform the bicarbonate contribution; carbonate, hydroxide and other bases may also contribute. Standard Methods 2340 defines total hardness and splits it into carbonate and non-carbonate parts [S25], and Standard Methods 2320 defines alkalinity as acid-neutralising capacity whose value depends on the chosen endpoint [S26] — a laboratory running the EDTA titration at pH 10.0 ± 0.1 with Eriochrome Black T or Calmagite and 0.01 M EDTA will find the step list in the certified-method checklist published by NC DEQ [S28]. In China, GB/T 5750.4-2023 covers organoleptic and physical indices and GB/T 5750.5-2023 covers inorganic nonmetallic indices [S29]; the National Public Service Platform for Standards currently exposes only the standard records, so confirm which part carries your parameter before citing a clause number.

pH decides how fast the film runs. In the 80 °C system used for the spiking series, lowering hard-water pH from 8.2 to 6.83 cut the surface carbonate deposit from 5.3 mg to 0.85 mg [S07] (index-level; that system). pH also decides something else: slightly alkaline liquors brown while EGC and EGCG decay [S13]. That alkaline decay does not establish a necessary catechin or flavour penalty from lowering pH. Measure where your liquor actually sits and measure catechin and flavour changes in the actual acidified formulation, rather than target a number borrowed from another system.

Boiling only reduces a film. “Boiling the water reduced but did not eliminate scum” is the reported result [S06]. Boiling also moves pH, and the direction is not the same everywhere: one study recorded the pH of London mains water falling from 8.0 to 6.67 after a one-second boil [S06], while a synthetic-water study recorded a rise of 0.5 to 1.0 units from released carbon dioxide [S13]. The net effect depends on the bicarbonate load and how freely the gas escapes, so record your own water’s pH before and after boiling and work from that pair.

Geometry and holding matter as much as composition. Film formation depends on the exposed surface area and is independent of the volume of tea: narrowing the vessel from 10.28 cm to 9.0 cm diameter lowered the formation rate by 19%, close to the 23% loss of area, while the mass after 60 minutes stayed unchanged [S06]. The reported activation energy of 34 kJ·mol⁻¹ explains why a preheated urn accumulating film for an hour outruns a cup that cools quickly [S06] (abstract/index-level). Any A/B comparison has to hold vessel geometry and fill constant, or the surface area will silently decide the result.

Water variable matrix: what each number predicts

Evidence grades: A = peer-reviewed full text or official primary record on file; B = abstract only; C = publisher abstract plus indexed passages; B+/A− = official-method sources graded on the same scale.

Variable	Effect on rim film	Effect on cold haze / sediment	Evidence
Total hardness (as CaCO₃)	Rises with hardness, but only where alkalinity is also present	Higher brewing-water Ca²⁺ shifts the liquor toward turbidity	[S01] A (artificial waters: none on Milli-Q, strongest at 200 mg CaCO₃/L); [S12] B
Alkalinity	In bicarbonate-dominated water, helps assess the bicarbonate half of the required pair; interpret with known composition and pH, not as a direct bicarbonate assay	No direct effect isolated in these sources — calcium in the bulk partners oxalate instead	[S06][S07] C (abstract/index level); [S12] B (oxalate pairing)
pH	Higher pH, faster film	Alkaline conditions drive browning and catechin loss rather than haze as such	[S07] C; [S13] A
Ca:Mg ratio	Substituting magnesium for calcium roughly halved film mass; doubling both raised it around 50%	Not tested against cream in these sources	[S06] C (that system’s magnitudes)
Heating and holding	Boiling reduced but did not eliminate (S06 system); longer warm holding and wider exposed area both raise the total	Cooling is the trigger — cream and sediment are described as two stages of one process	[S06] C (abstract); [S15] B
Chelator: citric acid	Weakens the film [S01]; lemon juice works through pH drop plus calcium complexation [S07]	Warming test tells you whether a chelator is even the right tool (reversible fraction only)	[S01] A; [S07] C
Chelator: EDTA	At 0.001 M no film formed after the infusion, and growth stopped in the following hour (pH fell alongside, so the two effects are entangled)	Na₂EDTA cut the reversible fraction by 14.6% and left the irreversible fraction untouched	[S06] C; [S14] B
Chelator: polyphosphates	Class-level evidence: polyphosphates sequester calcium in the infusion	Suppressed tannin–calcium turbidity when black tea was brewed with hard water	[S21] A (1962 study, polyphosphate class)
Milk / plant milk	Milk raises total surface solids yet suppresses normal film (47 mg vs 9.7 mg with 40 mL full-cream milk in 800 mL at 80 °C — most of the added mass comes from the milk itself, and the surface milk-fat layer limits oxygen access)	Oat milk tea shows a sediment layer on centrifugation; plant milk × rim film has no direct study	[S08] C; [S23] A
Fe³⁺ / Al³⁺	No direct film evidence — the aluminium in a film comes mainly from the leaf	Irresolvable cream carried more Al, Ca, Fe and Cu than resolvable cream, with Al³⁺ the strongest contributor to particle formation	[S05] C; [S24] B
Filtration / centrifugation	Ordinary clarification is not demonstrated here to prevent later film when dissolved precursors and holding conditions remain unchanged	Delays only: ≥4000 rpm × 20 min kept a concentrate stable for 21 days at 25 °C; ultrafiltered liquor turned cloudy again after one week in the refrigerator	[S09] C; [S18] C

One working day: the checks and their branches

Step	Setup	Observation	Reading	Next move
1. Parallel brews	Same tea lot, dose, temperature and time; identical vessel geometry and fill. Three arms: deionised water / current production water / current water plus a chelator	Film appears on the production-water arm only	The water side carries the fault	Go to the water matrix; fix alkalinity and pH before touching the powder
2. Warming test	Take the bottom-sediment sample and warm it gently to 40–60 °C	Turbidity falls	Reversible material — a chelator route is available for this fraction	Re-run with the chelator arm; the irreversible fraction will remain
3. Warming test (other branch)	Same sample	Turbidity holds	Irreversible material or insoluble particles	Look at oxalate-type anions, cooling profile and powder-side inputs — chelators will not recover this half
4. Acid test	Scrape and dry a piece of film, add dilute acid	Effervescence; the film thins and calcium-bearing particles dissolve [S05]	Carbonate is present in the film — the interface route is confirmed	Return to steps 1–2 to find which water ion supplied it
5. Optional grading	Centrifuge or filter the bulk sample in steps of force or pore size	How much stays suspended, how much redissolves on warming	Separates suspended particles, colloid and true solution	Treat as evidence about bulk particles, not proof of film prevention when dissolved precursors and holding conditions remain unchanged

Step 1 has two further exits: if all three arms grow film, water is not the discriminator — turn to geometry, holding and the detection limit of the method; if none of the three grows film, the phenomenon sits below that detection limit or is not ion-driven. Every step above has its own published support; the order itself is an engineering proposal. Evaporating a filtered sample to dryness to separate dissolved from insoluble solids follows Standard Methods 2540 [S27] — check the sub-section letter against your own copy of the method before writing it into a specification.

What to change first, and what each change costs

Water first. Water comparisons are a direct place to start when considering switching, blending or softening the supply: the deionised arm grows no film [S01]. Softening takes out calcium and magnesium; whether alkalinity falls with it depends on the process — precipitation softening removes carbonate alkalinity, ion exchange leaves it in place — so re-measure alkalinity after any treatment rather than assuming the hardness number describes the new water.
Chelator, chosen in three tiers. Citric acid works on two paths at once — pH drop and calcium complexation — and measurably weakens the film [S01][S07]. Polyphosphates are the next tier: a study on black tea infusion reported that they sequester calcium (without sequestering iron), suppress tannin–calcium turbidity in hard water, darken the liquor and raise extraction, with taste spoilt at 0.15% or above and acceptable at 0.1% in that system [S21] — a single 1962 study of the polyphosphate class, so treat the dose figures as that system’s conditions and confirm your own food-grade specification. The third tier is an enzyme route on the powder side: tannase lowered esterified catechins and cut green-tea concentrate sediment by 74.63% at 0.5 mg/mL over 120 min at 35 °C in its own system [S22]. EDTA is the most sensitive of the laboratory discriminators — it stops film growth [S06] and recovers the reversible sediment fraction [S14] — but any production use has to be checked against the rules of the destination market (US 21 CFR, China GB 2760, EU Regulation 1333/2008 Annex) before it enters a formula.
Heating and holding. Boiling reduces film load without removing it; holding temperature and exposed area then decide how much accumulates. Measure your water’s pH before and after boiling and set the holding window from that, not from a published direction.
Filtration and centrifugation, last among process options. The studies above report delayed bulk haze — 21 days after centrifugation, one week after ultrafiltration. Ordinary clarification is not demonstrated here to prevent later film when dissolved precursors and holding conditions remain unchanged. This does not cover all filtration or membrane treatments: treatments that change available calcium, bicarbonate or organic precursors may change subsequent film formation. Re-check the treated liquor rather than infer film prevention from bulk clarity.
Powder last. Caffeine is not required for film formation [S06], while it is a structural participant in cream [S04]. If the water arms are clean, move to the powder — and at that point the cream mechanism, not the film mechanism, is what you are troubleshooting.

The record to keep for each trial

Field	What to log
Water input	Total hardness as CaCO₃, alkalinity with its reporting basis, pH, Ca and Mg where available, sampling point and time — independent values, logged separately so each can be compared on its own
Trial arms	Water arm, chelator arm and dose basis, tea lot and dose, vessel geometry and fill (surface area is a variable)
Conditions	Brew temperature and time, holding temperature and duration, time points for each observation
Film result	Coverage or mass of surface film at the fixed time points, scored the same way across arms
Bulk result	Turbidity/appearance before and after gentle warming, sediment height or mass
Chemistry check	Acid test on the dried film; optional microscope look at the scraped material
Decision	Fault assigned to interface or bulk, next variable to change, date for re-check

Attach the water input block when you request samples for comparison: it turns a vague “our water seems to be the problem” into a comparable record. We supply different tea types and solids specifications per project and can run the water and process evaluation alongside your material — the checks above are yours to run on your own line, and the numbers they produce belong to your system.

Boundaries of these numbers

Every threshold in this article belongs to the system that produced it: Ca²⁺ above 40 mg/L came from a 1% green-tea infusion [S12], and 200 mg CaCO₃/L as the strongest film came from artificial water with one tea [S01]. Plant milk has been studied as a bulk-sediment system [S23] with no direct work on rim film; aluminium and iron evidence covers cream particles, not the surface film [S24]. The polyphosphate tier rests on one class-level study from 1962 [S21]. For colour changes during the same investigation, read Why Tea Drinks Turn Brown Without Turning Hazy.