GDL vs Calcium Sulfate for Tofu: Choosing a Coagulant Route and Qualifying the Change

For tofu manufacturers deciding whether to use an acid-set, mineral-salt, or combined coagulant system, the right comparison starts with the product and line—not a universal dose chart.

Related reading on GDL-set tofu

If the line already runs GDL and the problem is sourness or whey release rather than coagulant choice, start with diagnosing sour or weeping GDL silken tofu.

Start with the tofu you need to make

A supplier asks whether you want GDL or calcium sulfate. The useful answer depends on what the plant is making. A sealed, high-moisture silken tofu that sets in its retail pack has a different process target from a curd that must be cut, drained, pressed, and moved through a packaging line. Choosing only by price per kilogram skips the variables that determine whether the coagulant works in that system.

Glucono-delta-lactone (GDL, E575; CAS 90-80-2) and calcium sulfate (often called gypsum, E516) are both used in tofu production, but they do not set soy milk in the same way. GDL is an acid precursor: as it dissolves and converts to gluconic acid, the pH falls and soy proteins aggregate. Calcium sulfate supplies calcium ions that support protein aggregation through a mineral-salt route. The two ingredients may be used separately or in blends, but that does not make either one a drop-in replacement for the other [1, 2].

For a plant trial, first write down the target texture, package format, soy-milk solids, heating and hold profile, fill method, and whether the tofu will be pressed. Then compare routes against those conditions. A dose copied from a home recipe or another manufacturer is a starting question, not a production specification.

What the research does—and does not—show

A 2024 review of tofu coagulants groups GDL among acid coagulants and calcium sulfate among salt coagulants. It also surveys studies of blends. In one cited study, researchers evaluated yield, water content, protein content, sensory characteristics, and other measures for a GDL–gypsum combination. The review reports an optimum ratio for that experimental setup. That ratio belongs to that study's soy material and processing conditions; it is not a recommended formula for every tofu plant [1].

A 2022 paper examined calcium sulfate combined with GDL in tofu made from whole soybean flour. This is useful evidence that a combined system can be studied as its own route. It is not a direct factory trial on filtered soy milk, nor does it establish that a blend will improve yield or texture in a different recipe [2].

Another direct laboratory comparison examined GDL and magnesium chloride, not calcium sulfate. The authors held the other conditions constant and found similar changes in several measures of tofu-like precipitate formation, while discussing differences in how the precipitates formed. That study helps explain why coagulants should be compared under matched conditions; it cannot settle a GDL-versus-gypsum decision by itself [3].

This distinction matters because search results often compress the question to “Which coagulant makes the smoothest tofu?” A video or forum post can show how a home cook experienced one recipe. It cannot establish how a commercial line will behave with a different protein concentration, heat treatment, vessel, fill sequence, package, or cooling schedule. Use consumer discussion to identify the questions people ask—texture, sourness, dose, and mixing—not as production evidence.

GDL, calcium sulfate, or a blend?

The table describes questions for qualification, not guaranteed performance. “GDL gives soft tofu” and “gypsum makes firm tofu” are too broad to use as specifications. Soy-milk composition and thermal history influence the gel. Coagulant identity is one part of that system.

GDL may suit a process designed for a slow, distributed pH change, including some in-pack setting approaches. But the same gradual acidification that helps set a gel can move the flavor away from the target if the addition rate, concentration, or hold is wrong. A titration or pH check can help diagnose a trial, but pH alone does not tell you whether the finished tofu has the desired texture, water retention, or flavor.

Calcium sulfate may be chosen where the intended product and line already use a mineral coagulant. Its performance depends on how the powder is prepared and distributed. A plant should not assume a supplier's “tofu grade” title tells it the particle size, hydration form, dissolution behavior, or dispersion needed for its equipment. Ask for the controlled specification, test methods, and sample tied to a lot.

A blend is worth testing when a defined defect points to a controllable formulation problem. Do not start with a literature ratio and assume the result will scale. A blend changes more than one mechanism at once, so the trial needs a baseline, a clear reason for the change, and separate records for each component.

A changeover trial that can be interpreted

Keep the current commercial formula as the control. Make the smallest number of changes needed to answer the question. If the trial compares a GDL route with a calcium-sulfate route, hold the soy source, water ratio, solids target, heating profile, batch size, and packaging conditions constant. If those factors change too, a good or bad result will not identify the cause.

A false comparison often starts with dose basis. For each route, record whether the coagulant is added dry or as a slurry, the water used to prepare it, its concentration, addition point, mixing time, and the interval before filling. A GDL solution prepared in advance and a calcium-sulfate powder added directly to the tank are not just different ingredients; they are different preparation procedures. Ask what denominator the supplier uses for a suggested dose: soy-milk mass, total batch mass, dry solids, or finished product. Put that basis on the run sheet and make the bench method resemble the production sequence closely enough to expose mixing and fill constraints.

Agree on the test plan before weighing any powder. At minimum, record:

• Soy-milk total solids or the plant's defined protein/solids measure, with method and sampling point.
• Coagulant identity, lot, dose basis, and whether it was dry-added or pre-dissolved.
• Water used for any slurry, order of addition, mixing time, and mixer settings.
• Heating, hold, fill, seal, and cooling temperatures and times.
• Final package, fill weight, and time between filling and evaluation.
• A defined score for gel strength or texture, water separation, appearance, flavor, and yield.

Choose the measures that match the product. For in-pack silken tofu, visible whey and the integrity of the gel after the intended storage period may matter. For a cut and pressed product, curd handling, cut loss, press behavior, moisture, and downstream breakage may matter more. If the customer specification includes sensory limits, use the same method and trained panel conditions across the control and trial batches.

Run repeat batches rather than relying on a single tub. A one-off can expose a mixing problem; it cannot show normal process variation. If the result is close to a limit, test another lot before changing the production specification. Keep the results tied to coagulant lot numbers so a later purchase is traceable. Do not use pH as a substitute for food-safety validation. GDL and calcium sulfate are coagulants; a formulation change does not establish microbial safety, shelf life, or a kill step. Keep the plant's validated process and applicable food-safety plan under the responsible technical team's control.

Sourcing candidate for buyer review

One Alibaba listing for food-grade GDL (E575) is linked below as an RFQ starting point. It is a sourcing route, not a recommendation of one coagulant route over another — the changeover trial above still decides.

Normalize the quote after the technical fit is known

A quote comparison should use the amount of each ingredient required for an approved batch, not just a headline price per kilogram. But a conversion calculation cannot make two different processes equivalent. First qualify the route; then compare costs.

For a simple first-pass calculation:

Ingredient cost per batch = quoted cost per kg × kg used in the approved batch. Add packaging, freight, duties, sample/testing cost, and any new process expense separately. If a blend is selected, calculate each ingredient separately and record the combined cost. Do not compare a price for a 1 kg sample against a 20 kg or larger order tier as though the quote basis were the same.

Verify the material before you request a production sample

For GDL, ask the supplier to confirm the name, CAS 90-80-2, food-use grade, applicable specification, assay method and result basis, moisture, particle-size range if relevant, packaging, shelf life, and storage conditions. Get a representative COA and a controlled product specification. For a shipment decision, request the COA for the actual lot. A generic sample certificate is not lot evidence.

For calcium sulfate, pin down the chemical form and hydrate state rather than relying on “gypsum” or E516 alone. Ask for its CAS number, assay, moisture or loss-on-drying method, particle size, and any limits your process needs for insoluble matter or trace elements. Confirm that the document refers to food-grade material and to the manufacturing site supplying the order.

In the United States, 21 CFR 184.1318 lists glucono delta-lactone as a direct food substance affirmed as GRAS, and 21 CFR 184.1230 lists calcium sulfate. These entries do not replace an application-specific review, destination-market check, or compliance with good manufacturing practice. EU and other markets use their own permitted-use rules. Check the current rule for the actual product and jurisdiction before approving the ingredient [4, 5].

Record who manufactures and releases the lot, who invoices it, and how the supplier handles specification or site changes. A distributor can be suitable if the chain of custody and change control are clear. Do not infer manufacturer status from a marketplace product page.

A practical decision rule

Choose GDL when your target product and process are built around acid-set coagulation and a controlled fall in pH, and the trial meets flavor, texture, yield, and storage criteria. Choose calcium sulfate when the mineral-salt route fits the product and the line distributes it consistently. Test a blend only when it addresses a specific measured gap and you can control both components.

If none of the routes passes the same-batch comparison, stay with the qualified process and investigate soy solids, heat, addition, mixing, or fill conditions before buying a new coagulant. That is often cheaper than changing ingredients and still not knowing what caused the result. The decision is not “Which powder is best?” It is whether a defined lot and route reliably deliver the tofu your plant specifies, at the scale and market where it will be sold. Put that result in the change-control record; use the supplier's marketing description only as a lead for the next document request.

References

Geng, Y. et al. “Research Progress on Tofu Coagulants and Their Coagulation Mechanisms.” Foods. 2024;13(21):3475. https://doi.org/10.3390/foods13213475

Lu, W. et al. “Influence of calcium sulfate incorporated with gluconolactone coagulant on the quality of whole soybean flour tofu.” Food Chemistry: X. 2022;17:100527. https://doi.org/10.1016/j.fochx.2022.100527

Arii, Y., Sano, Y., and Nishizawa, K. “Direct comparison of the tofu-like precipitate formation by adding different coagulants: magnesium chloride and glucono-δ-lactone.” Heliyon. 2021;7(6):e07239. https://doi.org/10.1016/j.heliyon.2021.e07239

U.S. Electronic Code of Federal Regulations. 21 CFR §184.1318, Glucono delta-lactone. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-B/part-184/subpart-B/section-184.1318

U.S. Electronic Code of Federal Regulations. 21 CFR §184.1230, Calcium sulfate. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-B/part-184/subpart-B/section-184.1230