Reb D vs Reb M: One Glucose Unit Apart, Worlds Apart in the Formulation Lab

01 · Molecular structure

One Glucose Unit, Two Completely Different C-19 Chains

Both Reb D and Reb M share the same steviol aglycone — the ent-kaurene diterpene backbone common to all steviol glycosides. Both have an identical trisaccharide at C-13 (a β-glucosyl-1,2-β-glucosyl-1,3-β-glucopyranose branching pattern). The entire difference sits at C-19.

They are not isomers. The 162 Da mass difference corresponds to exactly one hexose (glucose) residue. Where Reb D terminates with a disaccharide at C-19, Reb M extends it by one more β-1,6-linked glucose to form a trisaccharide. That is the single structural change, and every property difference in this article flows from it.

Why the linkage type matters more than the glucose count

A 2025 study in Foods by Li and colleagues tested four steviol glycosides side by side — including Reb D and an α-1,6-glucosylated Reb A derivative — and measured solubility, melting point, fusion enthalpy and docking energy against the hT2R4 bitter receptor.^[1] Their key conclusion: glucosyl linkage type and position affect physical properties more than the raw number of glucose units.

Reb D’s sophorose unit at C-19 uses a β-1,2 linkage. This linkage constrains the sugar chain in a rigid conformation that hinders intermolecular hydrogen bonding with water molecules. The result is a molecule that packs tightly into crystals (high fusion enthalpy — 277.7 µVs/mg, the highest of the four glycosides tested) and dissolves reluctantly. Compare that to the α-1,6-glucosylated Reb A derivative, which has a flexible α-1,6 linkage: it achieves 137× the solubility of Reb D despite having the same number of glucose units.^[1]

02 · Sensory comparison

Both Beat Reb A — but Reb D Has an Onset Problem

Every competitor page on Reb D vs Reb M says the same thing: “both are less bitter than Reb A, both taste closer to sucrose.” That is true and also insufficient. When you look at the actual panel data, there is a specific, reproducible sensory gap between them that matters for formulation.

Prashad et al., 2020 — the 126-person panel

The most direct Reb D vs Reb M sensory comparison in the published literature comes from Prashad, Bhatt and Bhatt, published in Foods.^[2] They tested Reb A, Reb D and Reb M at 5 % and 10 % sucrose-equivalent concentrations in water, against sucrose controls, using a 126-person consumer panel. The results at the lower concentration (the one most relevant to beverage formulation):

The critical line is immediate bitterness. At the onset of tasting, Reb D’s bitterness was statistically significantly higher than sucrose (p < 0.05). Reb M’s was not (p = 0.27). Once the sip was in the mouth and the panellists assessed in-mouth bitterness, both glycosides scored around 1 — statistically identical to sucrose and far below Reb A’s 3.5. The difference is purely temporal: Reb D produces a brief bitter flash at the very beginning that Reb M does not.

No hedonic preference question was asked in this study, so we cannot say whether consumers noticed or cared about the onset difference. But in a formulation context — especially in clear, cold beverages where the first sip defines the product — a detectable bitter flash at onset is a reformulation trigger.

Korean descriptive panel (Food Sci Biotechnol, 2020)

A second study used 8 trained panellists with a 15-cm line scale (1.25 = “very fast,” 13.75 = “very slow” for onset; higher = more intense for attributes).^[3] Key numbers for Reb D:

Reb D’s sweetness onset was not significantly different from Reb M’s in this trained panel, and both were faster than Reb A. But the study confirms that Reb D still carries detectable bitterness (2.6), astringency (3.1) and licorice-type aftertaste (2.3) on absolute scales, even though all three sit well below Reb A’s levels. These are small enough numbers that an untrained consumer might not notice them in a blended formulation — but they are not zero.

CATA profiling (Prashad, 2020)

The same 126-person panel used Check-All-That-Apply profiling. Both Reb D and Reb M were checked for “artificial” flavour more often than sucrose. Both were checked for “bitter” less often than Reb A. Neither glycoside fully replicated the CATA profile of sucrose — but both were far closer to it than Reb A was.^[2]

03 · Solubility

The Practical Bottleneck: 430 ppm vs 1 000 ppm

Solubility is where the one-glucose difference has the most direct impact on formulation. If you cannot dissolve the glycoside, it crystallises during cold storage, hazes your product or settles to the bottom of the bottle. The published data:

Reb M is roughly 2.3× more soluble than Reb D in pure water at room temperature. Both are poorly soluble compared to Reb A (10.6× more soluble than Reb D). But the practical consequence is that Reb D hits its solubility ceiling at roughly 350–430 ppm, while Reb M stays comfortably in solution up to 1 000 ppm. For context, a typical usage level in flavoured water is 75–250 ppm, in carbonated beverages 150–400 ppm, and in dairy 150–450 ppm. Reb D is at the ragged edge of its solubility in carbonated and dairy formats. Reb M has headroom.

Reb D’s crystal polymorph problem

Urai, Kudo and Takiyama (2020) discovered that Reb D exists in two crystal polymorphs: Form α (a hydrate with 0.93 % water, phase transition near 50 °C) and Form β (anhydrous, 0.40 % water, phase transition near 70 °C).^[6] The practical relevance:

• Form α dissolves ~17 % faster than untreated Reb D at 7.5 °C — a modest but measurable improvement. This gain comes from the crystal lattice incorporating water molecules, which weakens intermolecular packing.
• Adding the hydrate form to a Reb D–saturated solution pushed dissolved Reb D above the apparent equilibrium for > 24 hours, suggesting a supersaturation window that could be exploited in cold-fill processes.
• All five commercial high-purity Reb D products the authors tested showed neither α nor β patterns out of the bag — the polymorphs only appeared after controlled drying at 50 °C or 100 °C. Manufacturers are not currently optimising crystal form.

No equivalent polymorph study has been published for Reb M. The US Patent 12 408 689 describes “Form A” and “Form B” crystal forms for Reb M, and notes that amorphous and spray-dried Reb M (0.3–1.2 % solubility) substantially outperform crystalline Reb M (0.1–0.15 %).^[10] But neither form has been characterised at the level of detail that Urai gave Reb D.

Solubility boosters

US Patent 9 060 537 documented several approaches to improve Reb D solubility:^[4]

• Potassium sorbate: pushed Reb D to ~4 000 ppm at 20 °C (a ~10× improvement). This works because sorbate disrupts the crystal lattice during dissolution.
• Glucose co-dissolution: improved Reb D solubility to ~8 000 ppm with spray-dried formulations.
• Spray drying: the most effective approach, converting crystalline Reb D to an amorphous or partially amorphous solid with dramatically better dissolution.

These fixes work, but they add cost and complexity. Reb M does not need them for any standard beverage application. For a deeper treatment of Reb M purity and its impact on solubility and haze, see our Reb M 95 vs 98 comparison.

04 · Acid stability

The Real Reason EverSweet Uses Both Glycosides

Most formulators assume Cargill blends Reb D and Reb M in EverSweet for taste synergy. The patent record tells a different story.

US Patent 11 812 767 (“Beverage having Reb D and Reb M”) discloses that Reb D degrades in aqueous beverages at low pH, producing impurities that hurt both taste and shelf life. The patent claims that adding Reb M to the formulation stabilises Reb D against this degradation:^[5]

The patent does not use the word “synergy.” The stated advantage is stability, not taste. At a 1 : 1 ratio in the pH 2.0–2.5 range, Reb D was most stable compared with Reb D alone.

What happens when steviol glycosides degrade at low pH

Prakash and colleagues (2014) characterised the degradation products of Reb M under acidic conditions (pH 2.0, 80 °C, 24 hours in phosphoric acid). They identified three degradation products:^[11]

• Compound 2: the C-16/C-17 exocyclic double bond migrates to C-15/C-16 — an alkene rearrangement within the steviol core.
• Compound 3: same rearrangement, plus hydration to a 16β-hydroxy group — a saturated ent-kaurane skeleton.
• Compound 4: conversion to the isosteviol skeleton (13-methyl-16-oxo-17-nor-ent-kaurane), losing three of six glucose units.

All three products were described as “minor,” and the study conditions (pH 2.0, 80 °C) are far more aggressive than any commercial beverage. Under normal shelf-life conditions (pH 2.8–3.2, 25 °C), degradation rates for steviol glycosides are slow — Reb A loses roughly 10 % after 113–182 days at pH 3, 40 °C.^[12] But the patent data suggest Reb D degrades faster than Reb M under the same low-pH conditions. If that holds in real beverage systems, the practical implication is clear: Reb D should not be the sole sweetener in any product at pH < 3.0 with a shelf life beyond 6 months.

05 · Production routes

Three Ways to Make Them — With Different Regulatory Consequences

Both Reb D and Reb M are available through three production routes, and the choice of route determines regulatory classification, labelling requirements and (in some markets) whether the product is legal at all.

1. Leaf extraction

Traditional water or ethanol extraction of stevia leaves, followed by resin purification and crystallisation. Because Reb D occurs at 0.4–0.5 % of leaf dry weight and Reb M at 0.02–0.2 %, commercial-scale leaf extraction requires massive amounts of biomass. PureCircle developed StarLeaf varieties bred for higher minor glycoside content, and GLG Life Tech used its Dream Sweetener breeding programme for the same goal.^[13] Leaf-extracted Reb D and Reb M can be labelled “stevia leaf extract” and are accepted in all markets.

2. Enzymatic bioconversion

Start with a purified steviol glycoside extract (typically ≥ 95 % total SGs, predominantly Reb A), then use UDP-glucosyltransferase (UGT) enzymes to add glucose units and convert lower glycosides into Reb D or Reb M. The enzymes are produced by recombinant Komagataella phaffii (formerly Pichia pastoris) or other expression systems.^[14]

A breakthrough came in January 2025 from RIKEN and collaborators: a single amino acid substitution in UGT91D2 — valine-155 to threonine (V155T) — increased Reb D and Reb M accumulation roughly tenfold in yeast and plant expression systems.^[7] This mutation enhanced UDP-glucose binding at the donor site and reduced side reactions with non-glucose donors. The practical significance: enzymatic production costs for Reb D and Reb M are falling, and the enzymatic route is likely to dominate new capacity additions.

Quality control for enzymatically produced Reb D

The enzymatic route introduces a regulatory concern that leaf extraction does not: residual recombinant DNA and residual host-cell protein from the expression organism. A 2022 EFSA dossier for an enzymatic Reb D process failed specifically on the DNA analytics, not the chemistry — the applicant could not demonstrate adequate clearance of recombinant DNA in the final product.^[17]

We test every batch of enzymatic Reb D for both:

• Residual recombinant DNA: limit of detection ≤ 10 ng DNA per gram of product.
• Residual protein: limit of detection ≤ 5 mg/kg.

These limits sit well below the thresholds set in current EU and US guidance. When evaluating enzymatic Reb D from any supplier, ask for these two data points on the COA. If they are absent, ask why.

Organic Reb D availability

Both RD95 and RD98 are available certified under EU Regulation 2018/848 and USDA NOP. Organic Reb D is chemically identical to conventional and performs the same in formulation. Certification is an access requirement for organic product lines, not a performance upgrade. Organic lots sometimes have longer lead times because they run in scheduled campaigns — if your timeline is tight, flag the organic requirement early. For the full range of organic steviol glycoside grades, see our stevia extract product page.

3. Whole-cell fermentation

Cargill’s EverSweet, produced by the Avansya joint venture with dsm-firmenich, uses the full biosynthetic pathway expressed in genetically modified yeast (Yarrowia lipolytica or Saccharomyces cerevisiae, depending on the generation). The yeast converts a simple sugar feedstock (dextrose from corn or sucrose from cane) directly into Reb D and Reb M without any stevia leaf input. Genes from stevia, rice, lettuce and tomato are expressed in the yeast.^[15]

Fermentation-derived steviol glycosides cannot be labelled “stevia leaf extract” in most markets, and some countries (notably some in Asia) restrict or have not yet approved bioconversion- or fermentation-derived stevia. GLG and ADM have positioned their leaf-extracted Reb M specifically for these markets.^[13]

06 · Regulatory landscape

11 GRAS Notices vs 4: Reb M Has a Three-Year Head Start

Both glycosides have FDA GRAS status. But the depth and breadth of regulatory coverage is not equal.

The numerical gap matters. Each GRN covers a specific production method and set of specifications. Reb M’s 11 GRNs span leaf extraction, fermentation via Saccharomyces cerevisiae, brewer’s yeast, Yarrowia lipolytica, and enzymatic conversion via E. coli and Pichia pastoris. Reb D’s 4 GRNs cover leaf extraction (PureCircle, GLG), fermentation via Saccharomyces (Blue California) and enzymatic conversion via Pichia (Sichuan Ingia). More GRNs means more production-route flexibility for formulators and fewer regulatory barriers when switching suppliers.

E 960a vs E 960c: the label split that matters

Within the EU’s E 960 framework, leaf-extracted steviol glycosides fall under E 960a and enzymatically or fermentation-derived glycosides under E 960c. The distinction controls what you can print on the label. E 960a products can be labelled “stevia leaf extract”; E 960c products cannot — they must declare “steviol glycosides” with the production method. For brands that position on “natural” or “from the leaf,” this difference is commercial, not just regulatory. We offer both routes for Reb D: leaf-extracted (E 960a, higher cost, “stevia leaf extract” label) and enzymatic (E 960c(iii), lower cost, “steviol glycosides” label). The choice depends on your target market’s label sensitivity.

The China gap

China’s new food raw material approval system has active Reb M applications on file, even though the pathway has been difficult (one application was disapproved in May 2023, a second entered public comment in October 2023). For Reb D, no filings appear in the public database. This means any China-facing formulation that requires steviol glycosides beyond the standard E960 leaf-extract specification currently has a regulatory pathway for Reb M but not for standalone Reb D. For a full Reb A vs Reb M comparison including regulatory coverage in Japan, Korea and ASEAN, see our companion guide.

07 · Cost and sweetness economics

Similar Potency, Slightly Different Price Points

Both glycosides have similar sweetness potency, but the published ranges vary across sources:

At face value, Reb D is about 12–15 % cheaper per kilogram than Reb M. But the cost comparison shifts once you factor in formulation complexity:

• Solubility management: Reb D may need spray drying, co-solubilisation or formulation aids to stay in solution in cold-fill or clear beverages. These add $5–15/kg to the effective raw material cost.
• Stability testing: In acidified products (pH < 3.0), Reb D requires more extensive accelerated stability testing and may need Reb M as a co-stabiliser, which adds the Reb M cost anyway.
• Regulatory filing: If your product sells into China or markets that have not yet approved fermentation- or enzyme-derived Reb D, the regulatory pathway cost can exceed the raw material savings many times over.

In practice, the ~$20/kg price difference between Reb D and Reb M disappears for most clear, acidified beverage applications once you account for solubility engineering and stability risk. It persists — and is a genuine saving — in opaque, neutral-pH formats (dairy, protein shakes, confectionery) where Reb D’s solubility limit is not tested and acid degradation is not a concern. It also persists in blends: replacing 10–25 % of the Reb M fraction with Reb D in a Reb M-dominant system captures most of the cost saving without hitting Reb D’s formulation constraints.

One additional cost factor: organic certification. Organic Reb D (USDA NOP / EU Organic) carries a premium over conventional, and organic lots run in scheduled campaigns with longer lead times. If your product line requires organic certification, factor the campaign schedule into your sourcing timeline — the raw material cost is secondary to missing a production window.

08 · Decision framework

When to Use Reb D, When to Use Reb M, When to Use Both

Use Reb M as the primary glycoside when:

• Your product is clear and cold-filled — flavoured water, clear RTD tea, transparent energy drinks. Reb M’s 2.3× solubility advantage means no crystallisation risk at standard usage levels.
• Your product is acidified below pH 3.0 — carbonated soft drinks, citrus beverages, kombucha. Reb M is more stable than Reb D at low pH, and the patent record shows Reb D needs Reb M as a co-stabiliser in this range.
• First-sip impression is critical — any product where the consumer evaluates the first sip (premium RTD beverages, single-serve formats). Reb M does not produce the brief bitter onset that Reb D does.
• You need China market access — Reb M has active new food raw material applications. Reb D has none.
• You want the broadest supplier flexibility — 11 GRAS filings across multiple production routes vs 4.

Reb D can work well when:

• The product is opaque, neutral-pH and served chilled — flavoured milk, protein shakes, ice cream, yoghurt. Solubility is not tested, acid degradation is not a factor, and the 12–15 % cost saving is real.
• The product is dry-mix or confectionery — solubility is irrelevant in a tableting or hard-candy application, and Reb D’s high crystal fusion enthalpy actually helps with tablet hardness and stability.
• You want the weakest bitter-receptor binding — Reb D’s −5.5 kcal/mol docking energy against hT2R4 was the lowest measured, consistent with very low in-mouth bitterness. In matrices where onset bitterness is masked by other ingredients (vanilla, chocolate, caramel), this advantage persists without the onset penalty.
• You are fixing a Reb A formula that already works but the sweetness onset is too slow or the finish lingers too long. Adding 10–30 % Reb D (as a share of total glycoside) tightens both. Returns diminish above 30 % — if the formula still needs work at that level, switch the primary glycoside to Reb M rather than pushing the Reb D fraction higher.

Blending ratios as starting points

These are screening ranges from our application work, not finished specifications. Convert on a Reb D basis, not extract weight — for RD95, multiply the target Reb D level by ~1.05. Dissolve the Reb D fraction hot and separately before adding it to the base, and hold the trial for a week at real storage temperature before accepting it.

Our bench experience on the D + M blend specifically: a 60 : 40 Reb M : Reb D blend at 250 ppm in citrus-flavoured water scored indistinguishably from pure Reb M in a 30-person internal screening. The $20/kg cost saving at that ratio is meaningful at scale — roughly $4–5/kg saving on the glycoside fraction, which compounds across high-volume beverage runs. Screen two ratios (we typically start at 30 % Reb D and work down) and pick the one where the onset bitterness stays undetectable in the matrix.

Use the D + M blend when:

• You are formulating a carbonated beverage at pH 2.0–2.8 and want to use Reb D for cost savings. The patent data shows Reb M at a 1 : 1 ratio with Reb D maximises Reb D stability.^[5]
• Your fermentation-derived ingredient is already a D + M mixture — as in EverSweet, where both are co-produced. Separating them adds cost; using the blend directly is the simplest path.
• You want taste synergy at lower cost — PureCircle’s application data suggests that glycoside blends “create taste synergies that reduce the need for masking agents.”^[13] The evidence is directional rather than quantified, but it aligns with the bench data above. At 10–25 % Reb D in a Reb M base, you capture most of the cost saving without touching the sensory ceiling.

09 · Side-by-side summary

Reb D vs Reb M: Complete Comparison

10 · Frequently asked questions

FAQ

References

References

Structure & physical properties

[1] Li J et al. Influence of the glucosyl linkage type on the aqueous solubility, physicochemical stability and sensory properties of steviol glycosides. Foods. 2025;14:1917. (Reb D solubility 0.43 mg/mL; fusion enthalpy 277.7 µVs/mg; hT2R4 docking −5.5 kcal/mol.)

Sensory evaluation

[2] Prashad H, Bhatt M, Bhatt S. Consumer-based sensory characterization of steviol glycosides (rebaudioside A, D, and M). Foods. 2020;9(8):1026. doi:10.3390/foods9081026 (n = 126; Reb D immediate bitterness p < 0.05 vs sucrose; Reb M p = 0.27.)

[3] Park S et al. Sweetness profiles of glycosylated rebaudioside A and its molecular properties. Food Sci Biotechnol. 2020;29(9):1309–1318. doi:10.1007/s10068-020-00873-w (8 trained panellists; Reb D onset 4.4, bitterness 2.6, astringency 3.1 on 0–15 scale.)

Solubility & crystal form

[4] Prakash I et al. Method to improve water solubility of rebaudioside D. US Patent 9 060 537. Filed 2012, granted 2015. (Reb D cold equilibrium 350 ppm at 23 °C; potassium sorbate → ~4 000 ppm; spray-dried → ~8 000 ppm.)

[6] Urai T, Kudo S, Takiyama H. Polymorphism of rebaudioside D and its effect on solubility. Food Sci Technol Res. 2020;26(1):17–23. doi:10.3136/fstr.26.17 (Form α hydrate, transition ~50 °C, 0.93 % water; Form β anhydrous, transition ~70 °C, 0.40 % water; Form α ~17 % solubility gain.)

Acid stability & degradation

[5] Prakash I, Bunders C, Engles S. Beverage having Reb D and Reb M. US Patent 11 812 767. Filed 2019, granted 2023. (Reb M stabilises Reb D at pH 2.0–2.5; 1 : 1 ratio optimal; 50–350 ppm each.)

[11] Prakash I et al. Structural characterization of the degradation products of a minor natural sweet diterpene glycoside rebaudioside M under acidic conditions. Int J Mol Sci. 2014;15(1):1014–1025. doi:10.3390/ijms15011014 (pH 2.0, 80 °C, 24 h; three degradation products including isosteviol derivative.)

[12] Gong Q. Storage stability of rebaudioside A in various buffer solutions. MS Thesis, Auburn University. 2013. (Reb A 10 % loss in 113–182 days at pH 3, 40 °C.)

Production & enzyme engineering

[7] Shoji T, Tanaka Y, Nakashima Y et al. Enhanced production of rebaudioside D and rebaudioside M through V155T substitution in the glycosyltransferase UGT91D2 from Stevia rebaudiana. J Agric Food Chem. 2025;73(3):2019–2032. doi:10.1021/acs.jafc.4c09392 (V155T → ~10× Reb D / Reb M accumulation.)

[14] UK Food Standards Agency. Safety assessment on a modification of specifications of rebaudioside D (E960c) to include manufacture by enzymatic conversion. RP1245. 2024. (UGT-a enzyme from K. phaffii; ≥ 95 % Reb D; 200–350× sweetness.)

Regulatory & commercial

[8] InVivoChem. Rebaudioside D — compound datasheet. 2024. (MW 1 129.15; leaf abundance 0.4–0.5 %; water solubility < 0.1 mg/mL.)

[9] Food Standards Australia New Zealand (FSANZ). Application A1108 — risk and technical assessment: rebaudioside M. 2019. (Crystalline Reb M solubility 0.1 g/100 mL at 25 °C; in vitro hydrolysis complete at 16 h.)

[10] Prakash I et al. Compositions and methods for improving rebaudioside M solubility. US Patent 12 408 689. Filed 2020, granted 2025. (Crystalline M + D mixture 0.1–0.15 %; spray-dried 0.3–1.2 %; Form A / Form B crystal forms.)

[13] Staff report. Minor stevia extracts like Reb M and Reb D are in high demand — how can stevia suppliers scale up supply? Nutritional Outlook. May 2018 / updated 2026. nutritionaloutlook.com

[15] Cargill. EverSweet stevia sweetener FAQs. cargill.com/food-beverage/na/eversweet-faqs (Fermented via engineered yeast; genes from stevia, rice, lettuce, tomato; sugar feedstock.)

[16] Icon Foods. Stevia ingredient pricing — screening estimates. 2024. (Reb D 95 ~$140/kg; Reb M 95 ~$160/kg; indicative, not contractual.)

Quality & safety

[17] EFSA Panel on Food Additives and Flavourings. Safety assessment of steviol glycosides (E 960c) produced enzymatically. 2022. (Applicant unable to demonstrate adequate clearance of recombinant DNA in enzymatic Reb D; dossier did not proceed.)