Hygroscopic, oily, and low-moisture fills. Which capsule material, and why

Which capsule material should I use for my fill?

HPMC for anything that pulls water or has to stay dry. Either shell, sealed, for a plain oil, with gelatin ahead where oxygen matters and HPMC ahead where the label says vegetarian. HPMC for any liquid that has glycerin, propylene glycol, ethanol, or a light PEG in it. Table 1 lists the fills that decide it, and the rest of the page explains each row.

Table 1. Which material, by what's in the capsule. Start with the row that describes your fill. If two rows point different ways, the one higher in the table usually wins, because a fill that damages the shell is a harder problem than a label or a price.

If you'd rather work through your own fill, the checker further down asks about the fill's traits and returns a shell, a sealing plan, the packaging, and the tests. Go to the checker. The AI prompt pack does the same for a whole formula, ingredient by ingredient. Get the AI prompt pack. And if you'd rather test than read, ask for a sample box of each material and run the trial in the second-to-last section. Each is one box at the single-box price, credited against your first order of that product, and four weeks and a box of each answers most of the questions on this page for your fill.

Why do gelatin capsules crack when I fill them with a hygroscopic powder?

Because the powder drinks the shell. A gelatin capsule leaves the factory holding 13 to 16% water, that water is what keeps the shell flexible, and a hygroscopic powder sitting inside it pulls the water out until the shell drops below roughly 10 to 12% and turns brittle. HPMC leaves the factory at 4 to 7% water and stays flexible even when dried to under 1%, so the same powder inside an HPMC shell has almost nothing to take and does no harm when it takes it.

Here's the mechanism in plain terms. Every material that holds water has a relationship between how much water it holds and the humidity of the air it's sitting in. The technical name is a sorption isotherm, and the number that matters is water activity, which is the humidity a material would create in a sealed space around itself. A gelatin shell at 14% water sits at a water activity of about 0.5, meaning it's in balance with air at 50% relative humidity. A hygroscopic powder as it comes out of the drum, at 1 to 3% water, might sit at a water activity of 0.1 or 0.2. Close the two together in a capsule and a bottle, and water moves from the higher activity to the lower one until they agree. The shell loses, the powder gains, and where they meet depends on how much of each there is and how thirsty the powder is. A 120 mg shell against a 700 mg fill of something that wants water is a fight the shell loses.

The threshold is better stated as a humidity than a percentage, because the sources agree on the humidity and not on the percentage. The classic paper on gelatin brittleness found that empty gelatin capsules become brittle once the surrounding humidity falls below about 40%, and later work on filled capsules found the same figure for the water activity the shell and its contents settle at. Below about 0.4, the shell cracks. In percentage terms, different sources put the onset anywhere from 7% to 13% water, and the makers' own line is "below about 10%." We use 10 to 12% and treat 0.4 as the number to design to.

The study that shows the whole thing happening is a six-month comparison of a senna dry extract in gelatin and in HPMC capsules, stored at 40 °C and 75% relative humidity. The gelatin shells started at 14.4% water and finished at 12.3 to 13.2%. By the end, 42 of 100 gelatin capsules had broken. The HPMC shells held at 4.4% water and 1 of 100 had broken. The extract inside the gelatin capsules had also lost most of its marker compound while the one inside HPMC had kept most of it, which is the other half of the story. When the shell gives up water, the fill takes it, and a fill that took on water is a different fill.

A textbook summary of the same effect puts it more bluntly. Gelatin capsules filled with a strongly hygroscopic material were "almost 100% broken" once the shell moisture dropped below 10%, while HPMC shells were undamaged at 2%. In 72 hours at 45 °C the gelatin shells lost 22% of their weight and the HPMC shells lost under 4%. The two shells are compared side by side on each property, and what a dry warehouse does to them is covered separately. This page is about what the fill does to the shell from the inside.

Two things follow from the mechanism. First, a desiccant in the bottle does to a gelatin shell what a hygroscopic fill does, on purpose. One maker's guidance puts it as "safe for HPMC but risks gelatin embrittlement," and the finished-product section of the storage page works through sizing a desiccant so it dries the fill and not the shell. Second, the shell alone doesn't protect a hygroscopic fill from a humid room. The published studies that cycled filled capsules through open humidity found that neither material kept the fill dry for long. The shell buys hours to days. The bottle, the seal, and the desiccant buy the shelf life. HPMC's job here is to survive being next to the fill, not to guard it.

How do I know if my ingredient is hygroscopic?

Weigh a sample, leave it in a sealed box at 80% humidity and 25 °C for 24 hours, and weigh it again. The European Pharmacopoeia's hygroscopicity test (chapter 5.11) is that simple, and the percentage weight gain puts the ingredient in one of four classes. Under 2% is slightly hygroscopic. From 2 to 15% is hygroscopic. Over 15% is very hygroscopic. And if the sample has turned to liquid, it's deliquescent. Under 0.2% isn't a pharmacopoeial class, but it's the practical line for "doesn't care."

You can run that test on a bench with a sealed plastic box, a saturated salt solution in the bottom to hold the humidity (potassium chloride gives about 84% at 25 °C, which is near enough for a screen), and a four-figure balance. If the ingredient came with a specification, the hygroscopicity class or a water activity figure is sometimes on it, and the maker's technical sheet often says outright when a material is deliquescent. If nothing is on paper, the 24-hour test costs an afternoon and answers the question for good.

The other number worth having is water activity, because it tells you which way water will move without having to reason about isotherms. A water activity meter reads it directly from a sample in a few minutes, and most contract manufacturers and any food-safety lab have one. Read the fill as it comes out of the drum. Below about 0.4, a gelatin shell is going to give up water to that fill and is at risk. Between about 0.4 and 0.6, the two are close to balance and gelatin can work. Above about 0.6, the fill is wet enough to soften a shell, and HPMC softens too if it goes high enough. That single reading is the most useful thing to have in front of you when you pick a shell, and it's the first thing the prompt pack asks for.

Table 2. Hygroscopicity classes and what they mean for the shell. Ph. Eur. 5.11 method, 25 °C, 80% RH, 24 hours. The shell column is our reading of the published mechanism, not a pharmacopoeial rule.

For scale, a published survey of raw materials by that method found betaine anhydrous gaining about 102% of its weight in 24 hours, povidone 27%, and lactose 0.5%. Mannitol and magnesium stearate gained about 0.2%. So a formula that's mostly betaine is a very different proposition from one that's mostly lactose, and the excipients you add to fix flow or bulk can move the whole blend's class. Measure the blend, not just the active, the same way you measure density for sizing.

Which ingredients cause trouble in capsules?

Four families cause most of the trouble. Salts and cholines that pull water until they liquefy. Extracts, hydrolysates, and fruit or honey powders that cake. Liquids that either wick out of the seam or dry the shell. And a small group of fills that react with gelatin rather than just wetting or drying it. Table 3 is the inventory we've been able to assemble from the literature, the makers' notes, and what formulators report, and the last column says which kind of evidence stands behind each row.

Table 3. Ingredients with reported capsule problems. Documented means a journal paper, a patent, a pharmacopoeial text, or a capsule or ingredient maker's technical data. Reported means a formulator, a pharmacist, or a retailer describing it without published data. We're measuring the reported ones on our own bench and will move them across as the data comes in.

Two things about that table. First, the "Reported" rows aren't less real, they're less measured. Alpha-GPC turning to goo is described by three different ingredient companies and by more formulators than I could count, and we couldn't find a published paper on it. Second, a handful of ingredients people ask about a lot have no data either way that we could find. MSM, glycine, electrolyte blends with sea salt, turmeric and curcumin extracts, green tea extract, berberine, and CoQ10 powders are all on our bench list, and the ingredient rows above will grow as the results come in. If you've had a capsule problem with an ingredient not on the list, reply to any email from us and tell us what happened. We'd like to add it.

The commonest fix, when the ingredient can't change, is HPMC plus a sealed pack with a desiccant. The section on fixing the fill covers what to do when the fill itself has to change, and the trial section covers how to prove either answer before a production run.

Which capsule for probiotics, enzymes, and other fills that have to stay dry?

HPMC, and for the most sensitive fills an extra-dry HPMC grade. The problem with a dry, moisture-sensitive fill runs the opposite way from a hygroscopic one. Here the fill isn't thirsty, it's fragile, and a gelatin shell at 14% water is the wettest thing in the bottle. Water moves from the shell into the fill, and a fill that had to stay at a water activity of 0.2 finds itself at 0.4.

Probiotics are the clearest case. Freeze-dried cultures are stable when their water activity is very low, and the published guidance from a contract manufacturer specializing in them puts the cliff at about 0.3. Above that, viability falls fast. A gelatin shell sits at roughly 0.5. The same manufacturer's reasoning is the one we'd give. The shell is a moisture source, HPMC is a smaller one, and the 10 to 100-fold overage that probiotic makers build into a formula is partly there to pay for moisture the shell and the pack let in. Enzymes behave the same way with their activity, and one capsule maker names enzymes as a target for its extra-dry HPMC, filled in a room at 15 to 25% humidity. Freeze-dried extracts, some hygroscopic APIs, and anything labeled "store in a dry place" on the drum belong in the same group. Nothing here is about what any of those ingredients do for a person. It's about keeping a dry powder dry.

The standard HPMC shell at 4 to 7% water is usually enough. The extra-dry grades, at 2 to 3.5% water, exist for fills where even that is too much, and they come with two conditions. They're shipped in sealed foil so they arrive at that moisture, and they have to be filled in a dry room, one maker says 15 to 25% relative humidity, or they take water back from the air before they reach the fill. A standard HPMC capsule in a normal 45% room will do more good than an extra-dry capsule filled in the wrong room. Gelatin has an extra-dry grade too, at 9 to 10.5% water, and at that moisture it's already near brittle, so it's a niche answer. We can supply extra-dry HPMC as a custom order, shipped sealed in foil, and the custom capsule page is where to start one.

The pack matters more than the shell over the shelf life. An induction-sealed bottle with a desiccant canister sized to the headspace, or a foil blister, is what holds a moisture-sensitive fill for two years. Sizing the desiccant, and the stability study that proves the pack, are worked through in full elsewhere. HPMC's part of the job is not to fight the desiccant, and it doesn't.

Can I fix a hygroscopic fill instead of changing the shell?

Sometimes, and the four routes are worth knowing even if you end up in HPMC anyway, because HPMC solves the shell and not the powder. A powder that cakes in the drum still cakes in an HPMC capsule.

The first route is an adsorbent. Silicon dioxide at 0.5 to 2% is the standard flow aid and does a little against moisture. A high-capacity carrier such as magnesium aluminometasilicate (sold as Neusilin) does a lot more, holding two to three times its own weight in liquid and turning an oily or wet-prone material back into a free-flowing powder. Two of the ingredient makers in the table above solve alpha-GPC that way, selling it as a 50% powder on silica. The cost is fill volume, because the carrier takes space, so the fill weight calculation has to be redone with the carrier in it.

The second route is granulation. Wet or dry granulating a hygroscopic blend reduces its surface area, slows its uptake, and fixes its flow and density at the same time. A contract manufacturer with a roller compactor will often reach for this before changing the shell, and for a light extract it can also solve a sizing problem. The trade-off is a process step and, for some actives, a stability question.

The third route is the one that surprises people. Pre-moisten the fill. If the fill is going to pull water from the shell until the two agree, bring the fill's water up to where they already agree and nothing moves. The paper that established this took a hygroscopic drug at 1.7% water, raised it to 7% before filling, and found that the brittleness of the gelatin shells over twelve months went away. The same logic sets the "balanced amount of water" for liquid fills, and the figure depends on the fill. It only works if the ingredient tolerates the water, which a probiotic or an enzyme does not, and it only works inside a sealed pack, because an open bottle undoes it. For a mineral salt or an amino acid in gelatin, it's sometimes the cheapest fix there is.

The fourth route is the room. Fill at 35 to 45% relative humidity, keep the drum closed, keep the hopper covered, and don't leave filled capsules on an open tray overnight. A hygroscopic powder measured on a wet Tuesday is a different powder from the one measured on a dry Thursday, and so is its density, which is where a lot of fill weight drift comes from. Fill weight drift on the line has its own troubleshooting entry.

The route we don't recommend is swapping to gelatin because the powder "runs better" and hoping the bottle will save it. The senna study above is what that looks like at month six.

Which fills react with gelatin, and what does cross-linking do?

Anything that carries an aldehyde, or makes one on storage, can cross-link a gelatin shell. The shell forms a thin, clear, water-insoluble skin, the capsule swells in the dissolution bath but won't let go of its fill, and once the reaction has started it doesn't stop when the cause is removed. HPMC doesn't do this at all.

The aldehyde sources are more ordinary than the word suggests. Polyethylene glycols oxidize in storage and form formaldehyde. Corn starch stabilized with hexamethylenetetramine releases formaldehyde and ammonia. Lactose and other reducing sugars react with amines in the fill. Chondroitin sulfate is on the list. So are some plant extracts, rayon coil in the bottle (furfural), and the metal ions in some colorants. Heat, humidity, and UV light all speed the reaction up, which is why a gelatin product can pass dissolution at release and fail at six months in the accelerated stability study. The US Pharmacopeia's dissolution chapter allows a second tier of testing with an enzyme in the medium to prove the failure was cross-linking and not the product, and how cross-linking and the enzyme tier work is set out with the gelatin and HPMC comparison. For a supplement with no dissolution specification, the symptom is a customer complaint about capsules found intact.

If the fill has a known aldehyde source and the product is gelatin, three things help. An antioxidant in a liquid fill slowed cross-linking in a six-month university study, where the PEG systems that contained tocopherol still dissolved fully at the end and the ones without didn't. A type B (lime-processed) gelatin cross-links more slowly than type A. And running the dissolution at three and six months on stability catches it before a customer does. The fourth option is HPMC, and for a formulator who already has a reason to prefer HPMC, an aldehyde source in the fill usually settles it.

One reactive fill has nothing to do with the shell and gets blamed on it anyway. Unstabilized R-alpha-lipoic acid polymerizes with heat and friction into a gum, sometimes inside an hour in a hot car according to its maker, and the result looks like a capsule that has gone wrong. The capsule didn't do anything. The fix is the stabilized sodium or potassium salt, and refrigerated storage for the free acid. Nothing here is about what the ingredient does for anyone, it's about a powder that turns to glue.

Can you put oil in a gelatin capsule?

Yes, and manufacturers have done it commercially since the 1980s. Both gelatin and HPMC two-piece capsules take oils, pastes, suspensions, and hot melts. The fill goes into the body to about 90% of its volume, the cap goes on, and the joint is sealed, either with a band of the same material around it or by fusing the cap to the body with a fine spray of water and alcohol and a little warmth. The trade calls the result a liquid-filled hard capsule, and a peppermint oil medicine sold in a banded size 1 gelatin capsule has been on pharmacy shelves in the UK for about 40 years.

The reason people think oil needs a softgel is that unsealed two-piece capsules leak, and most people's first experiment is an unsealed two-piece capsule. A soft gelatin capsule is made and sealed around its fill in one operation on a rotary die machine. A two-piece capsule is made empty, and if you fill it with a thin oil and snap it shut, the oil finds the gap between the cap and the body within days. The seal is what makes a two-piece capsule a liquid container. With the seal, the two-piece has some advantages a softgel doesn't. It's filled on equipment you may already own, in your own room, in small runs. It takes a fill that melts as high as 70 °C, where a softgel needs the fill below about 35 °C. It takes a tablet or pellets in oil in the same capsule. And it comes in HPMC, which puts a vegetarian label on an oil product without a softgel line. The softgel section further down sets the two against each other.

Some makers also sell a liquid-fill grade, a two-piece capsule with a tighter cap-to-body fit and a lock designed to take a band or a fused seal. We can supply liquid-fill grade capsules in gelatin and HPMC as a custom order, and the custom capsule page is where to start one. Whether the shell should be gelatin or HPMC for an oil turns on three things, the vehicle, the oxygen, and the label, and the next three sections take them in turn.

Which liquids can go in a gelatin capsule, and which need HPMC?

Plain oils and lipophilic excipients go in either. Water-loving liquids go in HPMC. The line runs between a fill that leaves the shell's water alone and a fill that pulls it out, and the published incompatibility list for gelatin is short and specific. Ethanol, glycerol (glycerin), propylene glycol, and polyethylene glycol below a molecular weight of about 4000 can't be used on their own in a gelatin shell. Medium-chain monoglycerides, sorbitan monooleate (Span 80), and diethylene glycol monoethyl ether (Transcutol) are on the same list. HPMC tolerates the glycols and the alcohols much better, and PEG from about 900 upward is fine in it, where gelatin needs 4000.

Table 4. Liquid vehicles against shell material. From the pharmaceutical literature on liquid-filled hard capsules, the excipient makers' data, and the capsule makers' technical notes. "Either" means both shells have been used commercially and the choice is made on oxygen and label. Confirm any fill in your own trial.

The one nuance the trade tends to skip is that the single head-to-head academic study of lipid fills, run in 2021 with five oils, five surfactants, and three cosolvents at 25 °C, found gelatin the more compatible shell overall for that set, and found that adding a little water to the formulation improved compatibility for both. So "HPMC tolerates everything better" isn't what the data says. HPMC tolerates the hydrophilic vehicles better. For a plain oil, gelatin is at least as good and brings a better oxygen barrier with it.

For any fill in the caution rows, the rule of thumb from the liquid-fill literature is the one from the storage page. If the filled capsule's moisture content doesn't move by more than 2% over six months at 25 °C and 65% relative humidity, a gelatin shell will keep its integrity. The trial section below is how to check it in four weeks rather than six months.

How much oil fits in a size 00 capsule, and in each other size?

About 0.82 ml in a size 00, which is 750 mg of an oil at 0.92 g/ml. The rule for a liquid fill is 90% of the body volume, because the last 10% is where trapped air goes and where the cap has to seat, and the makers and the pharmaceutical literature agree on the figure. Table 5 runs that arithmetic for every standard size.

Table 5. Liquid fill volume and weight by capsule size. Nominal capsule volumes as on the sizes page, filled to 90%. Fill weight is that volume multiplied by the oil's density. Most vegetable and fish oils sit between 0.90 and 0.93 g/ml; MCT is about 0.94. Read the column nearest yours.

So a 1,000 mg fish oil dose doesn't fit a single two-piece capsule of any size that most adults will swallow, which is one reason the market for that dose is softgels. A 500 mg oil dose fits a size 0. A 250 mg dose fits a size 3. A 300 mg dose in a size 2 at 0.92 g/ml is 98% of the liquid fill volume, which is tight; go to a size 1 or accept that the pump has to be set precisely. The makers publish the nominal volumes slightly differently from each other, as the capsule size chart explains, so treat the table as the place to start and the sample lot as the place to finish.

Two operating notes. Fill temperature stays below 70 °C in gelatin, and one maker quotes up to 80 °C for HPMC, because above that the shell softens. And after filling, hold the capsule upright for 10 to 100 seconds before closing so the air pressure from the fill can dissipate, or the cap lifts.

Why are my oil capsules leaking, and do I have to seal them?

They're leaking because a thin liquid wicks into the gap between the cap and the body, and yes, a fill that flows at room temperature has to be sealed. The industry line from one maker's sealing specialist is that any two-piece capsule without a seal "has a high likelihood of leakage," and the cut line of the body is where it starts. Thin oils are the worst offenders. MCT oil, the carrier a lot of small brands reach for, is thin enough that formulators report leaks starting within days in both gelatin and HPMC capsules and getting worse over weeks.

There are three ways to stop it, and they're not mutually exclusive. Thicken the fill, band the joint, or fuse the joint.

Thicken the fill. The pumping window for a liquid fill is roughly 100 to 1,000 centipoise (0.1 to 1 Pa·s) at the filling temperature, and the low end of that range is where leaks live. Water is about 1 cP and MCT oil is about 25 to 30, so MCT is well under the window. Blending in a heavier oil, a wax, or a thixotropic agent such as silica or beeswax brings the fill up, and the same additives that stop the leak also stop the fill settling in the pump. The further step is a semi-solid. A fill built on a glyceride that melts at 44 to 50 °C (the lauroyl and stearoyl polyoxylglycerides are the usual pharmaceutical choice) is pumped warm, sets in the capsule, and doesn't flow at room temperature, so it can't leak the way an oil does. For a bench or a compounding pharmacy, that's the route that avoids buying a bander. For a commercial product, the makers' guidance is to seal anyway, for tamper evidence and oxygen. For an over-the-counter drug in a two-piece hard gelatin capsule, the seal is required, because 21 CFR 211.132 says the capsule must be sealed using an acceptable tamper-evident technology in addition to the tamper-evident pack, and the FDA's compliance policy guide on tamper-resistant packaging (CPG 450.500) says sealed capsules alone are not tamper-resistant packages.

Band the joint. A band is a strip of gelatin or HPMC solution laid around the cap-body seam by a rotating wheel and dried. The band material matches the shell. A gelatin band on a gelatin capsule, an HPMC band on an HPMC capsule. That second one took the makers years to get right, because HPMC loses strength when it's wetted and a water-only HPMC band solution gave 14% leakers in one maker's patent data before they moved to a water-alcohol mix. Banding machines run from a bench unit at about 100 capsules an hour to production lines at 100,000 and more, and the band has to dry for hours. Banding is what the peppermint oil capsule uses.

Fuse the joint. The other route sprays a fine mist of water and ethanol into the seam, which lowers the melting point of the shell where it lands, then passes the capsule through warm air at 40 to 60 °C so the cap and body fuse. The maker that developed it quotes a larger sealed area than a band and no external band to inspect, and the fluid in the original patent is about 30% ethanol in water, not the 50/50 mix that gets repeated. Fused capsules harden overnight on trays. Integrated fill-and-seal machines for two-piece capsules run at 1,200 to 1,500 capsules an hour at the small end.

Whichever you choose, test it. The only quantitative leak test we could find in the open literature is from a sealing patent. Store the sealed capsules overnight, put them under a vacuum of 250 millibar for 20 minutes, and inspect on a light table for oily stains. The acceptance levels quoted were no more than 0.5% leakers at lab scale and 0.05% at production scale. Inverting the capsules on absorbent paper for 24 hours is the bench version most people use, and we couldn't find it written down as a standard. Either way, weigh 20 sealed capsules on day 0 and day 7. A leak is a weight loss, and so is a volatile fill escaping through a seal that looks fine.

One more cause of "leaking" that isn't a leak. A liquid fill in a gelatin capsule that has dried the shell will show a hairline crack at the shoulder before anything comes out, and an oil that has softened a shell will show the cap sliding on the body. Both are compatibility failures and belong to the trial section, not to the bander.

Which capsule material for fish oil and other fills that oxidize?

Gelatin, if the label allows it. Pullulan, if the label has to be vegan. HPMC, only with the rest of the oxygen defenses in place. Oxygen runs the other way from water. HPMC is the more forgiving shell for moisture and the more porous one for oxygen, and the published figure from one review of HPMC as a shell material puts the difference at about 50 times, roughly 3 cubic centimeters per square meter per day through gelatin against about 166 through HPMC. The material page uses the same figure and notes that the primary measurement behind it is hard to trace, so treat the multiple as directional and the direction as solid.

The seam matters more than the wall. One maker's sealing team states that about 98% of the gas that enters a two-piece capsule comes through the gap between cap and body, and that band-sealing a locking capsule cuts oxygen ingress about 60-fold. So an unsealed gelatin capsule is a worse oxygen barrier than a sealed HPMC one, and for an oil the seal is the first defense whatever the material. After the seal come the others. Flush the capsule with nitrogen at filling, which the integrated fill-and-seal machines do as standard. Put an antioxidant in the oil. Pack in a foil blister or an induction-sealed bottle with an oxygen absorber. And run a peroxide value on stability, because peroxide is how an oil tells you it's oxidizing before a customer smells it.

For a fish, krill, or flax oil where the label doesn't need to be vegetarian, gelatin plus a seal plus nitrogen is the combination the pharmaceutical liquid-fill literature is built on. Where the label does need to be vegan, pullulan is the shell the makers now sell for the job. Its makers claim an oxygen barrier several hundred times better than HPMC and several times better than gelatin, without publishing the absolute number, and it costs two to three times the price of gelatin. HPMC for an oxidizing oil is a legitimate choice when the vegan label is fixed and the price of pullulan isn't acceptable, on condition that the seal, the nitrogen, the antioxidant, and the barrier pack all do their jobs. Softgels, for what it's worth, contain 20 to 30% glycerin as a plasticizer, and plasticizers open channels for oxygen, which is part of why a sealed two-piece gelatin capsule can hold an oil at least as well as a softgel does.

Two-piece hard capsule or softgel for an oil?

A two-piece capsule when the run is small, the fill is in-house, the fill melts above about 35 °C, the capsule has to hold a tablet or pellets as well as the oil, or the label needs HPMC. A softgel when the oil dose is over about 500 mg per unit, the vehicle is hydrophilic, the volume justifies a contract softgel run, or the product has to look like a softgel because the category does.

Softgels are made on a rotary die machine that forms, fills, and seals the capsule from two ribbons of plasticized gelatin in one pass, and very few companies outside the specialists do it in-house. Contract softgel manufacturers quote minimums that trade sources put at 100,000 to 300,000 units per SKU as standard, with very few going below 50,000, and first runs at 8 to 16 weeks. Softgel filling rooms run at 20 to 30% humidity and the process wastes a third or more of the gelatin ribbon. Against that, a softgel holds more oil per unit than any swallowable two-piece, seals hermetically as part of being made, and takes hydrophilic fills that a two-piece gelatin shell can't.

A two-piece liquid fill is done on equipment a manufacturer often already owns, with a liquid dosing head and a bander or a fill-and-seal unit, in a room at up to 60% humidity, in runs of a few thousand if that's what's needed. The formulation work is done in-house with a few grams of fill, all the process steps happen in a day, and the shell comes in HPMC for a vegetarian claim. The fill can melt at 70 °C, which opens up wax and glyceride matrices that a softgel can't take. And the two-piece takes combination fills, a tablet in oil or pellets in oil, which a softgel can't. The peppermint oil medicine mentioned earlier and a prostate medicine that puts a tablet of one drug into an oil solution of another are both two-piece products for those reasons.

The cost comparison depends on volume more than anything. One capsule maker's position is that liquid-filled two-piece capsules come in at a comparable manufacturing cost to softgels and move faster from formulation to shelf, and for runs under about 100,000 units that's usually right, because the softgel minimum and lead time dominate. Above a few hundred thousand units of a 1,000 mg fish oil, the softgel line wins on volume per capsule and there's no two-piece answer to a dose that size anyway. What an order of empty capsules ties up in cash is worked through in the buying guide, if that's the comparison you're making.

What about oily or waxy powders like lecithin and CoQ10?

Fix the powder before you pick the shell, and then lean to HPMC. An oily powder in a two-piece capsule brings a different set of problems from an oil. It sticks to the tamping pins and the dosing disc, it gives wandering fill weights because its density changes as it warms, the oil migrates into the shell over months and shows as a translucent stain, and in a gelatin shell the oil can soften the wall where it touches. We couldn't find a published oil percentage for a powder blend above which those things start, so we won't invent one. The practitioner reports say the problems start well below the point where the powder looks wet.

The fix is an adsorbent carrier. Colloidal silicon dioxide at 1 to 3% helps a mildly oily powder flow and takes up some free oil. For a powder that's carrying a lot of oil, lecithin and oil-loaded extracts especially, the high-capacity carriers do the work. Magnesium aluminometasilicate (Neusilin) holds 200 to 330% of its own weight in oil and stays a free-flowing powder, and its maker's data shows it doesn't exude the oil back out under compression, which is what a tamping pin does to it. The trade-off is volume. A carrier at 30% of the blend is 30% more capsule, and the sizing arithmetic has to be redone with the carrier in it (what to do when the dose won't fit).

Once the powder is fixed, HPMC has two things in its favor. It has less water for the oil to displace, and it has no plasticizer in the wall for the oil to migrate into. CoQ10 powders and curcumin extracts, both of which are oily, low-density, and colored, are on our bench list because the reported staining is real and the data isn't. If the fill can't be fixed as a powder, the other answer is to make it a liquid on purpose, dissolve or suspend it in an oil and fill it as a liquid, which turns an awkward powder into an ordinary liquid-fill job.

Can you put essential oils in capsules, and will they dissolve the shell?

Yes, sealed and diluted, and no, a properly diluted essential oil in a sealed two-piece capsule doesn't dissolve the shell. The precedent is a peppermint oil medicine sold since the 1980s as 0.2 ml of peppermint oil diluted in arachis (peanut) oil in a banded size 1 gelatin capsule. The pattern is the pattern for any volatile fill. Dilute the essential oil in a fixed oil, fill to 90%, band or fuse the joint, and treat weight loss on stability as the number that matters, because a volatile fill escapes through a marginal seal as vapor long before it drips as a liquid.

What the retail advice gets wrong is the neat oil. The consumer guides that tell people to drop peppermint or oregano oil into an empty capsule and "swallow it immediately" are describing a capsule that's about to fail, because neat terpenes and phenols attack both gelatin and HPMC at the cut line, and because an unsealed capsule of a thin liquid leaks. Dilution and a seal fix both. We couldn't find a peer-reviewed compatibility study of limonene, garlic oil, or oregano oil against either shell material, so those three are on our bench list rather than in a table, and until the data exists the safe reading is the pharmaceutical one. Dilute, seal, and weigh on stability.

Nothing here is about what an essential oil does for anyone. It's about keeping a volatile liquid inside a shell until it's swallowed.

How do I test my fill against a shell before a production run?

Fill 60 capsules in each candidate shell, seal them in bottles, hold them four weeks at room conditions and four weeks at 40 °C and 75% humidity, and measure the shell's moisture, the fill's moisture, and how many capsules break or leak at each pull. Four weeks in the accelerated leg tells you most of what a six-month study will, and it's the difference between finding out on your bench and finding out from a customer.

Here's the protocol we'll run on your fill, and the one the prompt pack writes out for you. First, characterize the fill as received. Loss on drying or Karl Fischer for moisture, water activity with a meter, and the 24-hour hygroscopicity test from earlier on this page. Those three numbers predict most of what follows. Second, fill 60 capsules in each shell you're considering, on the equipment you'll use in production or as close as you can get, at the fill weight you'll use. Weigh 20 filled and 20 empty from each lot. Third, split each lot across two induction-sealed bottles of 30 with no desiccant, and put one bottle at 25 °C and 60% humidity and one at 40 °C and 75%. If your pack will have a desiccant, run a third bottle with it. Fourth, pull at 7, 14, and 28 days. At each pull, weigh 10 capsules, then separate shell from fill on 5 and run loss on drying on each. Count cracks under a lamp, then drop the 10 from 30 cm onto a steel plate and count breaks. For a liquid fill, do the 250 millibar leak test or the inverted-paper test on 20 and count stains. Fifth, at 28 days, run disintegration in water at 37 °C on 6 from each bottle and compare with day 0.

The pass criteria are the published ones. Gelatin shell moisture stays above 12% and the drop test breaks none of 10. HPMC shell moisture moves by less than 2 percentage points. Fill moisture moves by less than 2 percentage points in either shell (that's the six-month rule of thumb applied early). No leakers above 0.5% for a sealed liquid. Disintegration still inside 30 minutes at 28 days, and if gelatin has slowed markedly, the enzyme tier of the dissolution test tells you whether cross-linking has started. Fail any of those in the accelerated leg and the answer is the other shell, a fix to the fill, or a different pack, and you found out for the price of two boxes of capsules and a month.

If you'd rather we ran it, send us the fill. We fill it into gelatin and HPMC at the size you're considering, run the four-week protocol above, and send you the shell moisture, the fill moisture, and the breakage or leak counts at each pull, laid out so you can put them straight into your quality file. Email us first, through the contact page, with what the fill is, the size, and the weight you're aiming for. Don't post anything until you've heard back, because we need to know what's coming before it arrives. We'll reply with the address of whichever of our offices is nearer, UK or US, and the sample goes there marked for sample testing.

The full written protocol, with the saturated-salt jars for the humidity points and the record sheet, is what the prompt pack produces for your own fill, so your quality team can file it as a study. Run it again when you change capsule supplier, because a new maker's shell can sit at a different moisture from the one you qualified, and where the trial box fits when you switch supplier is covered with the rest of the change.

The fill compatibility checker

Describe your fill and the checker returns a shell, whether it needs sealing, what the pack needs, and the tests to run before production. Pick the fill type first, because the questions change with it. For a powder it asks about hygroscopicity class or water activity, moisture sensitivity, oil content, and aldehyde sources. For a liquid it asks about the vehicle, the viscosity, whether the fill flows at room temperature, and the dose in milligrams so it can pick a size from Table 5. For both it asks whether the fill oxidizes and what the label has to say.

It works from the rules on this page, and it says which rule it applied and where the figure came from. What it can't do is ask you a follow-up question, or screen the six ingredients in your blend one at a time and tell you which one is the problem. The AI prompt pack does, because it asks. Get the AI prompt pack.

What is an AI prompt pack?

An AI prompt pack is a small set of files you load into the AI assistant you already use. The assistant then asks you the right questions about your fill, works through them with you using our reference data, and shows its working as it goes. We've done the prompt engineering so you don't have to. The prompt carries the rules, the thresholds, the ingredient inventory, and the trial protocol from this page, written so that ChatGPT, Claude, Gemini, or Grok applies them in order and says which rule it used at each step. We publish one with each article.

Inside the Capsule Fill Compatibility AI Prompt Pack is the prompt file, a workbook the assistant fills in as you go, and a short guide. Paste the prompt in and it asks for your ingredients one at a time, with their doses and whatever you know about each one's moisture, hygroscopicity, oil content, and reactivity. Where you don't know, it tells you the test to run and offers a provisional class from the inventory on this page, marked as an assumption. It asks about the whole fill (powder, liquid, or semi-solid, and the vehicle if it's a liquid), then about oxygen, then about the label, the market, the filling room, and the pack you plan to sell in. It gives you back a shell material with the reasons, the risk flag for each ingredient and the fix for each flagged one, a sealing plan and a size from the liquid table if the fill is a liquid, the packaging the fill needs, the four-week compatibility trial written out as a protocol with a record sheet, a stability plan, and the six questions to put to whichever capsule supplier you're talking to. Then it fills the workbook so your quality team has the screen, the trial, and the results in one file.

Why a pack and not just the checker? Because the checker can't ask whether the third ingredient in your blend is on the deliquescent list, or notice that your flow aid is corn starch and your shell is gelatin. The pack does, and it explains why it's asking. It works for the founder who has a formula from a contract manufacturer and has been asked "gelatin or veggie?" without knowing what's in the question. It works for the formulation scientist who knows the answer for the active and wants the excipients screened in ten minutes. It's the conversation we'd have with you on a call, packaged so you can have it at your desk with your formula in front of you.

Use the form below to request this resource. Newsletter updates are a separate, optional choice.

Get the Capsule Fill Compatibility AI Prompt Pack

If we haven't answered your question, or your fill doesn't fit a row in the table, reply to any email from us or write to us through the contact page. A person who has had a hygroscopic powder crack a box of gelatin capsules will get back to you. And if you'd rather run the trial than read about it, ask for a sample box of each material.

Related.

• Gelatin vs HPMC capsules. Which to use, and why
• Capsule moisture, storage, and shelf life
• Capsule sizes, volumes, and fill weights
• Capsule fill weight from bulk and tapped density
• Faults and troubleshooting on the capsule filling line
• Vegan, halal, kosher, and allergen claims for capsules

Sources.

• Gelatin 13 to 16% and HPMC 2 to 6% water, HPMC brittleness unaffected at low humidity: Ku M. S. and colleagues, International Journal of Pharmaceutics 386 (2010). https://www.sciencedirect.com/science/article/abs/pii/S0378517309007765 Gelatin brittle below 10%, "none of the capsule types protected their contents": Jones B. E., Tablets and Capsules. https://www.tabletscapsules.com/3641-Technical-Articles/598122-HPMC-Capsules-and-Hygroscopic-Fills-Some-Clarification/
• Brittleness of empty gelatin capsules below 40% RH, and the sorption-desorption moisture transfer model: Kontny M. J. and Mulski C. A., International Journal of Pharmaceutics 54 (1989). https://www.sciencedirect.com/science/article/abs/pii/0378517389901683 Water activity 0.4 threshold, GAB isotherm parameters, pre-moistening the fill from 1.7% to 7%: Chang R.-K. and colleagues, Journal of Pharmaceutical Sciences 87 (1998). https://doi.org/10.1021/js9704238
• Senna extract, 42 of 100 gelatin broken against 1 of 100 HPMC, shell moisture figures: Indian Journal of Pharmaceutical Sciences. https://www.ijpsonline.com/articles/a-comparative-study-of-hard-gelatin-and-hypromellosecapsules-containing-a-dry-extract-of-senna-cassiaangustifolia-under-controlled-4002.html Gelatin "almost 100% broken" below 10%, 22% against 4% weight loss in 72 hours: Analytical testing and evaluation of capsules, textbook chapter. https://basicmedicalkey.com/analytical-testing-and-evaluation-of-capsules/
• Desiccant "safe for HPMC but risks gelatin embrittlement": capsule maker's technical article, 2017. https://qualicaps.com/-/media/Project/Articles/HPMC/2017_Can-you-fill-HPMC-capsules-with-hygroscopic-formulations.ashx Extra-dry grades, 2.0 to 3.5% HPMC and 9.0 to 10.5% gelatin, filling at 15 to 25% RH, enzymes as a target: Coulman S. and colleagues, ONdrugDelivery, 2020. https://ondrugdelivery.com/evaluating-the-mechanical-properties-of-extra-dry-hard-shell-capsules/ and https://ondrugdelivery.com/quali-v-extra-dry-a-novel-capsule-for-delivering-hygroscopic-pharmaceutical-drugs/
• HPMC dried to under 1% without losing strength, storage windows: Processing magazine. https://www.processingmagazine.com/material-handling-dry-wet/article/21128487/the-perfect-climate-for-capsule-storage Vcaps Plus 6 to 7% water, working window 10 to 70% RH, no cross-linking: capsule maker's FAQ. https://www.capsugel.com/knowledge-center/nutra/capsules/vcaps-plus-frequently-asked-questions
• Hygroscopic and deliquescent fills embrittle gelatin, the 10% at 43% RH cut-off: US patent 5,037,698. https://patents.google.com/patent/US5037698
• Ph. Eur. 5.11 hygroscopicity classes and the 80% RH 24 hour method; betaine 101.9%, povidone 27.2%, lactose 0.5% at 80% RH: Pharmaceutical Development and Technology, 2022. https://www.pharmaexcipients.com/wp-content/uploads/2022/06/Data-driven-approach-to-mitigate-quality-impact-of-hygroscopic-pharmaceutical-raw-materials-throughout-the-supply-chain.pdf Hygroscopicity classification by water vapor sorption, peer-reviewed: https://pubmed.ncbi.nlm.nih.gov/21981708/
• Probiotic viability cliff at water activity 0.3, shell as moisture source: contract manufacturer's technical article. https://eubioco.eu/en/probiotics-in-dietary-supplements-how-the-choice-of-capsule-shell-and-lyophilizate-form-affects-cfu-survival-2/
• Deliquescence of vitamin salts and deliquescence lowering in blends: Hiatt A. N., Taylor L. S., Mauer L. J., International Journal of Food Properties. https://www.tandfonline.com/doi/full/10.1080/10942911003650338 Alpha-GPC liquefying at 50 to 60% RH: ingredient maker's data. https://purelife-bio.com/products/alphagrain/ Coated hygroscopic actives review (calcium chloride, ranitidine, pyridostigmine, extracts, fruit powders): Pharmaceutics 14 (2022) 2015. https://www.mdpi.com/1999-4923/14/10/2015 Glucosamine salts: US patent 4,642,340. https://patents.google.com/patent/US4642340A/en R-alpha-lipoic acid polymerization: ingredient maker. https://geronova.com/the-sticky-truth/ Efflorescent and deliquescent compounding powders: https://pharmacyinfoline.com/efflorescent-hygroscopic-deliquescent-powders/
• Gelatin cross-linking mechanism, aldehyde sources, USP <711> two-tier enzymes: Gray V. and colleagues, Dissolution Technologies, November 2014, and Lu X. and Shah P., Dissolution Technologies, August 2017. https://dissolutiontech.com/DTresour/201411Articles/DT201411_A01.pdf and https://dissolutiontech.com/issues/201708/DT201708_A01.pdf USP <711> Dissolution, Pharmacopeial Discussion Group harmonization page: https://www.usp.org/harmonization-standards/pdg/general-methods/dissolution Mitigation, type B gelatin and antioxidants: ONdrugDelivery. https://ondrugdelivery.com/strategies-to-mitigate-gelatin-cross-linking-in-oral-solid-drug-delivery/ Antioxidants and Gelucire 44/14 in six-month gelatin stress study: Kreiner and colleagues, University of Strathclyde, APS 2016. https://strathprints.strath.ac.uk/57632/1/Kreiner_etal_APS2016_Liquid_filled_hard_gelatin_capsules.pdf
• Gelatin incompatibility list (ethanol, glycerol, propylene glycol, PEG below 4000, Capmul MCM, Span 80, Transcutol), PEG 900 floor for HPMC, 70 °C limit, 0.1 to 25 Pa·s, 90% fill, the 2% over six months rule: Biyani M., Pharmaceutical Technology. https://www.pharmtech.com/view/selecting-excipients-liquid-filled-hard-capsules
• Gelatin more compatible than HPMC for a lipid, surfactant, and cosolvent set, water improving both: Pharmaceutical Research 38 (2021). https://link.springer.com/article/10.1007/s11095-021-03088-8 Balanced amount of water, 10 to 12% for PEG 400/PVP and 5 to 6% for Labrasol/Aerosil: Kuentz M. and Röthlisberger D., International Journal of Pharmaceutics 236 (2002). https://www.sciencedirect.com/science/article/abs/pii/S0378517302000224 Glycerin-based HPMC liquid fill stable 2.5 to 75% RH: patent WO2007029098. https://patents.google.com/patent/WO2007029098A1/en
• Ideal viscosity 100 to 1,000 cP, 90% body volume, softgel production figures: capsule maker in Tablets and Capsules, 2016. https://www.qualicaps.com/-/media/Project/Articles/Others/2016Tablets-and-capsulesFormulating-with-liquidstwopieces-or-softgel-capsules.ashx 80 °C for HPMC, 60% RH filling room, softgel comparison: https://blog.lyfegroup.com/why-liquid-filled-hard-capsules-may-be-a-better-option-than-softgels
• Fill to 90% and hold upright 10 to 100 seconds: US patent application 2002/0081330. https://patents.justia.com/patent/20020081330
• "High likelihood of leakage," 98% of gas through the seam, 60-fold reduction with banding, band material matches shell, bander throughputs: Nutritional Outlook. https://www.nutritionaloutlook.com/view/locked-and-loaded HPMC band solutions and the 14% leakers with water only: US patent 9,579,290. https://patents.google.com/patent/US9579290
• Fusion sealing, about 30% ethanol, 40 to 60 °C: US patent 4,756,902 and application note. https://patents.google.com/patent/US4756902 and https://www.pharmaceuticalonline.com/doc/cfs-1200-sealing-process-0001 Sealing comparison and nitrogen: https://www.capsugel.com/knowledge-center/capsules/liquid-filled-hard-capsules-sealing-technology
• Vacuum leak test, 250 mbar for 20 minutes, 0.5% and 0.05% acceptance: patent application CA 3202044. https://patents.google.com/patent/CA3202044A1/en
• Oxygen permeability 3.14 against 166 cc/m²/day; semi-solids up to 80 °C in HPMC: Majee S. B., Avlani D., Biswas G. R., International Journal of Pharmacy and Pharmaceutical Sciences 9 (2017). https://journals.innovareacademics.in/index.php/ijpps/article/view/20707 HPMC "less oxygen protection" and gelatin liquid-fill grade for air-sensitive fills: https://blog.lyfegroup.com/capsules-for-oil-based-formulas-capscanada
• Softgel minimums and lead times: https://sourcify.com/softgel-manufacturing/ Softgel plasticizer 20 to 30% and oxygen channels, sealing mandatory when the fill flows at ambient, 80 to 80,000 cP at fill temperature: Tablets and Capsules. https://www.tabletscapsules.com/3641-Technical-Articles/592486-Capsule-Filling-Solving-Formulation-Problems-with-Liquid-Filled-Hard-Capsules/ Liquid-filled hard capsules review: Cole E. T., Cadé D., Benameur H., Advanced Drug Delivery Reviews 60 (2008). https://www.sciencedirect.com/science/article/abs/pii/S0169409X07003171
• Gelucire melting points and HLB: https://www.pharmaexcipients.com/news/the-gelucire-family-semi-solid-excipients-by-gattefosse/ and Gattefossé, Gelucire 44/14 product page: https://www.gattefosse.com/pharmaceuticals/product-finder/gelucire-4414 TPGS melting point 37 to 41 °C: https://www.tpgs.com/tpgs-technical-info/tpgs-properties Thixotropic fills origin: Walker S. E. and colleagues, Journal of Pharmacy and Pharmacology 32 (1980). https://academic.oup.com/jpp/article-abstract/32/1/389/6198647
• Neusilin oil adsorption 200 to 330%: https://tech-en.fujichemical.co.jp/neusilin-oil/
• Peppermint oil, 0.2 ml in arachis oil, banded size 1 gelatin: Colpermin SmPC, electronic medicines compendium. https://www.medicines.org.uk/emc/product/1076/smpc
• MCT leakage reports: https://future4200.com/t/mct-only-capsules-leaking/737 and https://atg-pharma.medium.com/two-piece-hard-capsules-how-to-prevent-leakage-a551254b424
• 21 CFR 211.132, tamper-evident packaging for OTC drugs, two-piece hard gelatin capsules must be sealed using an acceptable tamper-evident technology: https://www.ecfr.gov/current/title-21/chapter-I/subchapter-C/part-211/subpart-G/section-211.132 FDA CPG 450.500 tamper-resistant packaging, sealed capsules alone are not tamper-resistant packages: https://www.fda.gov/regulatory-information/search-fda-guidance-documents/cpg-sec-450500-tamper-resistant-packaging-requirements-certain-over-counter-human-drug-products 21 CFR 111.70 specifications: https://www.ecfr.gov/current/title-21/chapter-I/subchapter-B/part-111/subpart-E/section-111.70
• Capsule volumes and shell weights: as compiled on the sizes page from the major makers' published charts.

*Capsules.com supplies empty hard capsules in gelatin and HPMC to manufacturers, brands, pharmacies, and research teams. Capsules Direct Ltd, company number 17362490, 26 Rowood House, Bicester OX26 4PP, United Kingdom. UK 020 3905 1989 · US (816) 445-0042 · capsules.com. *