Reb A vs Reb D vs Reb M: The Solubility Problem Nobody Puts on the Spec Sheet

01 Framing

The Batch That Passed Every Test and Failed Anyway

Assay confirms the glycoside is present at the stated purity. Sensory confirms it tastes right. Neither test asks whether the material will still be in solution in six weeks.

What the failure actually looks like

Crystallisation out of a steviol glycoside solution does not present as one symptom, and the form it takes is diagnostic. Before assuming the sweetener is the cause, work out which of these you have.

The last row is worth acting on before anything else. Redissolution behaviour is the cheapest available test for whether you are looking at a sweetener problem at all, and it takes a beaker and a water bath.

The rest of this page is about the five rows above it.

02 Sensory

The Sensory Upgrade Is Real, and It Is Not a Trade-Off

Start by conceding the point that motivates the switch, because it is well supported.

A consumer study at Michigan State compared Reb A, Reb D and Reb M against a 14 percent sucrose reference. In-mouth sweetness and bitterness for Reb D and Reb M were not significantly different from sucrose. Reb A was significantly more bitter than sucrose, than Reb D and than Reb M. A follow-up study applied the same three glycosides in ice cream at 0.09 percent w/v against a 14 percent sucrose control and again found the sensory advantage carried into a real matrix.

So the upgrade is not marketing. If bitterness is your limiting defect, moving up the glycoside range is a legitimate answer, and we have written separately about how to work out whether bitterness is actually your limiting defect.

Why this is not a trade-off, and why that matters

It is tempting to frame what follows as a bargain: cleaner taste bought at the price of poorer solubility, pick your compromise. That framing is wrong, and it is worth killing early, because everything in section 10 depends on it being wrong.

If taste and solubility were genuinely coupled, no molecule could have both, and the only available strategy would be downstream mitigation. But two independent 2025 studies each produced a modified steviol glycoside with both properties at once.

Both are better tasting than Reb A. Both are far more soluble than Reb A. There is no compromise being made in either case.

That distinction decides what kind of answer is available to you. A genuine trade-off could only ever be managed. Two independent properties can be engineered apart, which is what section 10 is about.

03 Structure

Reb A vs Reb D vs Reb M: More Glucose Does Not Mean More Soluble

The intuitive model is that steviol glycosides with more attached sugar units should be more hydrophilic and therefore more water soluble. Reb D carries one more glucosyl group than Reb A. By that reasoning it should dissolve better.

It dissolves roughly ten times worse.

The two isomer pairs introduced in section 02 are what settle the question, because in an isomer pair the sugar count is held constant and only the bond changes. Same atoms. Same molecular weight. Two orders of magnitude apart in water.

The 2025 Jiangnan University group that produced RA1G concluded from solvation free energy calculations that glucosyl linkage type and position matter more than glucosyl number. The mechanism is packing rather than polarity: how a molecule sits in its crystal lattice, and how easily water can get between the molecules and pull them apart, is decided by bond geometry and molecular flexibility rather than by how many hydroxyl groups you can count on paper.

If you want one sentence to settle any argument about predicting solubility from structure, the 2019 crystallography work on Reb B provides it. Every solubility prediction software package the authors tested failed to predict these compounds' solubility accurately, and failed even to rank them in the correct order.

04 Definitions

The Number on Your Datasheet May Not Be a Solubility

Published figures for the same glycoside vary widely, and most of the variation is not error. Different sources are answering different questions.

The patent literature is unusually precise about this, and the distinction it draws is the one formulators need.

• Instantaneous solubility. How much dissolves when you stir it into water at room temperature and look at it. This is what most quick screening tests measure, and it is the number that decides whether your batching step works.
• Equilibrium solubility. How much stays in solution once the system has settled. One published protocol defines it as heating to 80 °C for fifteen minutes, cooling to 25 °C, then observing for up to four days, with a clear solution free of precipitate counting as soluble. This is the number that decides whether your product survives its shelf life.

Under that protocol, Reb D shows an instantaneous solubility below 0.08 percent and an equilibrium solubility of 0.08 percent. Reb M shows an instantaneous solubility of about 0.13 percent and an equilibrium solubility of about 0.2 percent.

Note what those two pairs tell you. For Reb M, heating and settling roughly doubles what stays in solution. For Reb D, it changes almost nothing. The two glycosides respond to the same process differently, which means a process adjustment that rescues one will not rescue the other.

The four solubility questions a formulator is actually asking

Those two scientific definitions map onto four practical questions, and a single figure on a datasheet can only ever answer one of them. Knowing which one you need is what turns a supplier conversation from a request for a number into a request for the right number.

Most published and datasheet figures answer the first question. The fourth is the one that costs money when it is wrong, and it is the one nobody quotes.

05 Solid state

One Molecule, a Forty-Fold Range

This is the section that changes how a specification should be written.

Set the published room-temperature water solubility figures side by side, and add the figures for the same molecules in different solid forms. The scale below is logarithmic, because the values span more than three orders of magnitude and a linear axis cannot show them together.

Now compare the two kinds of variation, because this is where the usual Reb A vs Reb D vs Reb M discussion goes wrong.

The gap between glycosides, Reb A at 0.51 percent against Reb D at 0.05 percent, is about ten-fold. The gap within a single glycoside, Reb D from 0.05 percent crystalline to 2.0 percent in a formulated amorphous composition, is up to forty-fold. Reb A itself swings five-fold between its hydrate and non-hydrate forms.

Which leads directly to the conclusion this article exists to make. A specification line reading "Rebaudioside D 95 percent" tells you the chemistry and tells you almost nothing about the solubility. Two lots at identical assay, from different crystallisation or drying routes, can differ by several times in how much stays in your syrup. Assay is a composition control. Solid state is a performance control, and most stevia specifications do not have one.

One thing we are not going to claim

It would be commercially convenient to say that fermentation-derived or enzymatically produced Reb M is more soluble than leaf-extracted Reb M. The evidence does not support that as a general rule. Production route determines regulatory identity and impurity profile. It does not reliably determine solid-state behaviour, which depends on the crystallisation and drying steps at the end of the process rather than on where the molecule came from. Ask about the finishing steps, not the origin story.

06 Potency

What Sweetness Potency Buys Back

Low solubility matters less for a sweetener that works at 300 times the potency of sugar than it would for a bulk ingredient. So the fair way to compare these materials is not by solubility but by how sweet a saturated solution can be.

Multiply each solubility figure by its sweetness potency and you get a ceiling: the maximum sucrose equivalence a saturated aqueous solution of that material can deliver. The dashed line marks 10 percent sucrose equivalence, the sweetness of an ordinary soft drink.

Two readings of that chart matter.

Reb M's solubility is not a problem in most finished beverages. A 78 percent sucrose equivalence ceiling is many times what any ready-to-drink product needs. If someone is selling you a solubilised Reb M system for a finished RTD, ask them to show the arithmetic that says you need it.

Reb D's ceiling sits at the sweetness level of an ordinary soft drink. An 11 percent ceiling against a typical 10 percent sucrose equivalence target leaves essentially no headroom, in water, before you have concentrated anything at all.

These ceilings describe sweetness only. They say nothing about the dissolved solids, body and mouthfeel that sugar also provided, which is a separate problem with a separate answer and is covered in our guide to how sweetness and bulk get replaced by different ingredients.

07 Concentration factor

Where a Safe Finished Dose Becomes a Risky Syrup

Almost every discussion of stevia solubility talks about finished product concentration. Manufacturing rarely works at finished product concentration.

A 5x syrup, a 10x concentrate, a dosing premix or a liquid drop product all multiply the sweetener concentration by the same factor they multiply everything else. The arithmetic is trivial and almost nobody runs it. Each sloped line below is what a formula needs; each flat line is what the material can hold. The crossing point is where the process fails.

Reb D is the row that stands out. At an ordinary soft drink sweetness, in water, the finished product itself is already sitting on the reported solubility limit. No concentration step is required to reach the wall. Whether it holds in practice depends entirely on what else is in the formula, which is another way of saying it depends on things your supplier cannot know and you have to test.

The Reb M rows show the more common industrial situation. Comfortable in the bottle, marginal in the syrup room. This is why a product can pass every finished-product stability trial and still generate complaints from a co-packer running a concentrate.

Patent literature puts the practical industrial requirement at at least 0.3 percent w/w for syrup and beverage formulations. Crystalline Reb M at roughly 0.1 percent falls three-fold short of that, and crystalline Reb D falls six-fold short. That gap, not the taste, is what the solubilisation industry exists to close.

08 Process

Heating Solves Dissolution, Not Shelf Stability

Heat will get crystals into solution. It will not necessarily keep them there.

What heating produces above the equilibrium concentration is a supersaturated solution, which is thermodynamically metastable. It can look completely clear for hours or days and then nucleate, often triggered by a temperature drop, a seed crystal, agitation or a container surface. The failure appears in a warehouse or a cold chain, weeks after the QC release.

Two consequences for how you test.

• Checking clarity while the mix is hot proves nothing. The published protocols specifically heat, cool, and then observe over days, because that sequence is what separates a genuinely dissolved system from a supersaturated one.
• The relevant cold temperature is your customer's, not your lab's. A 25 °C laboratory is warmer than a winter warehouse, a refrigerated display and most shipping containers on a night crossing. Test at the coldest temperature the product will actually see.

One further detail worth knowing about the improved systems on the market. Where solubilised or disordered crystalline compositions are described in the patent literature, their claimed performance is often stated in hours and days rather than months. A composition described as remaining dissolved at 0.3 to 0.4 percent for one day or longer is a real improvement over crystalline material, and it is not the same claim as shelf-life stability. Read those claims for their time window, not just their concentration.

09 Two stabilities

Chemical Stability and Physical Stability Are Different Questions

These get conflated constantly, and the conflation is why a compliant certificate of analysis can accompany a failed product.

Steviol glycosides are thermally robust. Recent work found that stevioside, Reb M and glucosylated Reb M analogues all retained more than 92 percent of starting concentration after treatment at 120 °C, which is why these materials survive pasteurisation, retort and baking without difficulty. That robustness is real, and it is routinely quoted as though it settled the stability question. It settles one of the two.

When a customer reports sediment, the first instinct is usually to pull the assay. The assay will come back fine. It was always going to come back fine.

10 Route out

The α-Glycosidic Route, and What It Does and Does Not Promise

Section 02 established that taste quality and solubility are not coupled, and section 03 showed why: changing a single glycosidic bond from β to α moved solubility by 80-fold in one isomer pair and 137-fold in another, while sensory quality held or improved. That is not an academic curiosity. It is the chemistry underneath glucosylated steviol glycosides, designated E960d in the European Union and produced commercially by enzymatic transglucosylation using sucrose as the glucosyl donor.

Recent work on dual-enzyme synthesised Reb M analogues shows the mechanism directly. X-ray diffraction gave a relative crystallinity index of 34.4 percent for unmodified Reb M, while the glucosylated analogues fell below 0.5 percent, showing broad amorphous halos instead of sharp diffraction peaks. Glucosylation disrupts the ordered crystal lattice, and a molecule that cannot pack tightly is a molecule that dissolves more readily. The same study found no loss of thermal stability.

So this is a genuine structural solution rather than a formulation workaround, and it addresses the cause identified in section 05 rather than compensating for it downstream.

Other routes exist and are worth evaluating on their merits. Amorphous or spray-dried forms of Reb D and Reb M, disordered crystalline compositions, co-solute systems, and Reb M and Reb B blends have all been described. What they share is that they act on solid state or on the solution environment rather than on the molecule, so their benefit is generally time-limited in a way a structural modification is not.

11 Protocol

A Screening Sequence Before You Approve a Lot

This is a practical screening sequence for incoming material, not a regulatory or compendial test method. It is written to catch the failures described above at the sample stage rather than at the customer.

• Prepare at your highest real process concentration, not your finished concentration. If you run a 5x syrup, screen at 5x. The finished dose is not the test that matters.
• Use your actual matrix as well as water. Water tells you about the raw material. Your matrix tells you about your product. Run both, because the difference between them is itself useful information.
• Match your real process temperature, hold time and agitation. Then record how long dissolution took, not merely whether it happened.
• Cool to the coldest temperature the product will see, including transport and retail display, not laboratory ambient.
• Observe at 2 hours, 24 hours, 72 hours and 7 days. The 2 hour reading catches gross failure. The 72 hour and 7 day readings catch supersaturation, which is the one that reaches customers.
• Record clarity, haze, sediment, wall and ring deposits, and whether anything that came out redissolves on warming. Redissolution behaviour distinguishes crystallisation from other kinds of sediment, as set out in section 01.
• Repeat on a second lot before you write the specification. Lot-to-lot solid-state variation is the failure this whole exercise is designed to find, and one lot cannot show it to you.

Photograph each observation point against a dark background. Fine crystalline haze is difficult to describe in an email and unmistakable in an image, and a photographic record settles supplier discussions quickly.

12 Specification

What a Solubility Specification Has to Say

We will start with our own material, because it makes the point better than a hypothetical.

Our technical datasheets currently carry the line "soluble in water." That is standard across this industry and it is not good enough. It states no concentration, no temperature, no time to equilibrium, no solid state and no observation window, which means it cannot be used to predict anything and cannot be tested against. We are revising it, and in the meantime any customer who needs a measured value under stated conditions should ask us for one rather than relying on that line.

A solubility line that can actually be used answers all of the following.

• At what concentration, expressed as percent w/w or g/L rather than as a qualitative word.
• At what temperature, both the dissolution temperature and the temperature the result is claimed to hold at.
• Over what time, and specifically which of the four questions in section 04 the figure answers.
• Heated or unheated, and if heated, at what temperature and for how long.
• In what medium, since water and your matrix will differ.
• In what solid state, crystalline, amorphous, hydrate, spray-dried or co-processed. This is the line almost nobody asks for and the one section 05 says matters most.
• Particle size distribution, which governs how fast material dissolves even where it does not change how much eventually will.
• The maximum concentration factor the supplier will stand behind, which converts all of the above into an answer you can apply to your process.

The assay lines on a stevia specification remain necessary and we have written about them separately, including which glycoside the percentage actually refers to. The grades and assay bases we hold are listed on the stevia extract product page.

13 Cost

What the Premium Grades Cost Once Potency Is Counted

Comparing steviol glycosides by price per kilogram is meaningless, because they are not used at the same rate. The number that matters is cost per unit of sweetness delivered, and potency claws back part of the price gap before you get there.

The arithmetic is simple. Cost in use per litre equals price per kilogram multiplied by grams needed per litre, divided by one thousand. Section 07 already produced the second term.

Reb M is commonly quoted at several times the price of Reb A per kilogram, with a rule of thumb in the region of ten times, though that ratio moves with leaf harvests, fermentation capacity and forward contracting and should be checked against live quotations rather than taken from an article. Apply the potency correction and a ten-fold price premium becomes roughly an eight-fold cost-in-use premium. That is a real reduction and it does not change the order of magnitude.

What you are actually buying

Here is where the cost question meets section 06, and the answer is uncomfortable for anyone selling premium grades.

In a finished ready-to-drink beverage, Reb A has roughly thirteen-fold solubility headroom and Reb M roughly eight-fold. Neither is remotely constrained. So in that application the premium buys sensory quality and nothing else. It is not buying solubility, because Reb A is the more soluble of the two.

That is a defensible purchase if bitterness is your limiting defect and the category is premium enough to carry it. It is not defensible as a general upgrade, and it is worth stating the per-litre figure explicitly in a business case rather than letting a per-kilogram number carry the argument in either direction.

Its sensory profile genuinely is better than Reb A. Its cost in use is higher on both counts, price per kilogram and grams required. And its solubility is roughly ten times lower. In a high-solids matrix that combination can still be the right trade, because co-solutes carry the solubility problem and the sensory gain is real. In a clear low-solids liquid it rarely is, and we sell the material.

One further cost that does not appear on any quotation. A grade that fails on physical stability costs a held batch, a rework, a customer investigation and sometimes an account. Weighed against that, the screening sequence in section 11 costs an afternoon and some beakers.

14 Decision framework

Reb A vs Reb D vs Reb M: Which One Should You Evaluate First

Reb A vs Reb D vs Reb M is not a ranking, because there is not one. The right starting point depends on where in your process the sweetener has to survive.

Three things we will say against our own catalogue

• We sell Reb D, and in a clear beverage or a liquid concentrate it is frequently the wrong choice. Its sensory profile is excellent and its solubility ceiling in water sits at roughly the sweetness of an ordinary soft drink. If that is your application, we will tell you before you order rather than after.
• Reb M does not need solubilising for most finished beverages. Anyone selling you an enhanced-solubility system for a standard RTD should be asked to show the calculation that says you need it.
• Every figure on this page comes from published sources that disagree with each other, in some cases by a factor of several. Do not copy our numbers into your specification. Use them to ask your supplier, including us, for a measured value under stated conditions.

We supply Reb A, Reb D and Reb M grades, glucosylated steviol glycosides, and blended systems. Tell us the application, your highest process concentration factor, and the coldest temperature the product will see, and we will tell you which grades can hold it and send enough material to run the sequence in section 11 in your own matrix. Samples are free, MOQ is 1 kg, and quotations are returned within 24 hours. Full grade details are on the stevia extract page.

15 Questions

Frequently Asked Questions

These come up most often when buyers compare Reb A vs Reb D vs Reb M on physical behaviour rather than on taste.

16 Evidence base

References

[1] Zhou, Z., Wang, W., Guo, Q., Wang, H., Xia, Y. (2025). Structure and property relevance of two pairs of isomeric steviol rebaudiosides and the underlying mechanism. Foods, 14(11), 1917. DOI: 10.3390/foods14111917

[2] Comparative study of rebaudioside A and rebaudioside Aα: physicochemical properties and in vivo metabolism. npj Science of Food (2025). DOI: 10.1038/s41538-025-00585-7

[3] Characterization of a new hemihydrate rebaudioside B crystal having lower aqueous solubility. Food Chemistry (2019). DOI: 10.1016/j.foodchem.2019.125533

[4] Tao, R., Cho, S. (2020). Consumer-based sensory characterization of steviol glycosides (rebaudioside A, D, and M). Foods, 9(8), 1026. DOI: 10.3390/foods9081026

[5] Muenprasitivej, N., Tao, R., Nardone, S. J., Cho, S. (2022). The effect of steviol glycosides on sensory properties and acceptability of ice cream. Foods, 11(12), 1745. DOI: 10.3390/foods11121745

[6] Dual-enzyme biocatalytic synthesis of novel rebaudioside M analogs with enhanced aqueous solubility, physicochemical stability, and sensory profiles (2026). View source

[7] US Patent 11,407,780. Steviol glycoside compounds, compositions for oral ingestion or use, and method for enhancing steviol glycoside solubility. Defines instantaneous and equilibrium solubility and the associated test protocol. View source

[8] US Patent 11,291,232. Highly soluble rebaudioside D. Reports crystalline and amorphous Reb D solubility figures. View source