Anhydride Curing Agent Dosage: phr, AEW and Accelerators

Why Anhydride Doses Are Set by Equivalents, Not Percentages

Anhydride curing is a stoichiometric reaction: one anhydride group reacts with one epoxy group, in the presence of heat and usually an accelerator. That makes the dose a ratio of equivalents, not a fixed percentage of the batch, and it explains why two formulators using the same resin and the same anhydride can end up with different recipes that are both correct for their own conditions.

This article is the calculation side of the anhydride curing family. It sets out the formula, works through the arithmetic with real numbers, covers the accelerators and the cure profile, and lists the errors that show up most often in a plant that has just changed resin or anhydride supplier.

The Formula

phr = (AEW / EEW) x 100 x r

Symbol Meaning Where it comes from
AEW Anhydride equivalent weight, g per anhydride group Molecular weight divided by the number of anhydride groups
EEW Epoxy equivalent weight, g per epoxy group The resin certificate, or a titration
r Anhydride-to-epoxy molar ratio A formulation choice, normally 0.85 to 1.0
phr Parts of anhydride per hundred parts of resin, by weight The number that goes on the batch record

Anhydride equivalent weights in this family

Anhydride CAS Molecular weight Anhydride groups AEW
1,2,3,6-THPA 85-43-8 152.15 1 152.15
3,4,5,6-THPA 2426-02-0 152.15 1 152.15
HHPA 85-42-7 154.16 1 154.16
MTHPA, 41.0% anhydride group min 11070-44-3 166.17 1 166.18 theoretical, 175.7 at 41.0%
Nadic anhydride (NA) 826-62-0 164.16 1 164.16
MHHPA, 98% assay 25550-51-0 168.19 1 168.19 theoretical, 171.6 at 98%
4-MHHPA, 99% assay 19438-60-9 168.19 1 168.19 theoretical, 169.9 at 99%
MNA (methyl nadic anhydride), 98% assay 25134-21-8 178.18 1 178.18 theoretical, 181.8 at 98%
Succinic anhydride, 99.5% assay 108-30-5 100.07 1 100.07 theoretical, 100.6 at 99.5%
OSA (octenylsuccinic), 99% 42482-06-4 210.27 1 210.27 theoretical, 212.4 at 99%
NSA (nonenylsuccinic), 99% 28928-97-4 224.30 1 224.30 theoretical, 226.6 at 99%
DDSA (dodecenylsuccinic), 99% 26544-38-7 266.38 1 266.38 theoretical, 269.1 at 99%
DSA (dodecylsuccinic, saturated), 99% not assigned as branched mixture; straight chain 2561-85-5 268.39 1 268.39 theoretical, 271.1 at 99%

All thirteen carry a single anhydride group, so the equivalent weight equals the molecular weight. That is a convenience of this family rather than a rule: a dianhydride would have an AEW of half its molecular weight, and a formulator who carries a habit across product families without checking the number of reactive groups will be out by a factor of two.

The last two rows are there because they are the extremes of the same family. Succinic anhydride is 72% anhydride group by mass, so a kilogram carries 9.94 equivalents and the dose at EEW 185 and r = 0.90 is under 50 phr; MNA is the heaviest of the liquid hardeners at 181.8 g/eq and 5.50 equivalents per kilogram. Both are dosed with the same formula and neither can be dosed from its molecular weight alone – see MNA (CAS 25134-21-8) and succinic anhydride (CAS 108-30-5).

The alkenylsuccinic branch of the family reaches the same table from the other direction: instead of stopping at the ring, it hangs a hydrocarbon chain off it. OSA (212.4), NSA (226.6), DDSA (269.1) and its saturated twin DSA (271.1) all carry exactly one anhydride group and are dosed with the same formula – what makes them different is that up to 73% of a DDSA drum is flexibiliser rather than hardener. The chain-length logic is set out in the alkenylsuccinic anhydride family.

MTHPA is the one entry where the two columns of that table part company, and the reason is worth keeping. Its molecular weight is 166.17, but a specification of 41.0% anhydride group minimum describes a material that is only 94.6% anhydride, so the figure a dose calculation should use is 166.17 / 0.946 = 175.7. Calculating from 166.18 understates the requirement by about 5%, and the step from the solid grades is larger still: at EEW 185 and r = 0.90 the dose moves from 74.0 phr to 85.4 phr, which is 15%, not the 1.3% that separates THPA from HHPA. The grade-by-grade detail is in MTHPA grades.

Why r is less than one

The anhydride ring does not open by itself. It is opened by a hydroxyl group, either one present in the resin or one generated as the reaction proceeds, and each ring opening produces a new carboxyl that can react with another epoxy. Because hydroxyls participate in the network, a stoichiometric ratio of 1.0 leaves unreacted anhydride in the cured part rather than leaving unreacted epoxy. The practical ratio for casting systems sits between 0.85 and 0.90; some formulations run to 0.95 or 1.0 for specific property targets, and a few run below 0.85 where a softer, more flexible network is wanted.

Resin EEW r = 0.80 r = 0.85 r = 0.90 r = 1.00
170 71.6 76.1 80.6 89.5
185 65.8 69.9 74.0 82.2
200 60.9 64.7 68.5 76.1
250 48.7 51.7 54.8 60.9

The table is for THPA and 3,4,5,6-THPA (AEW 152.15). For HHPA, multiply every figure by 154.16/152.15, which is 1.013: at EEW 185 and r = 0.85 the result is 70.8 phr instead of 69.9.

Worked Examples

Example 1: a straight casting batch

Resin: standard DGEBA with a certified EEW of 185. Anhydride: THPA. Ratio: r = 0.90.

phr = 152.15 / 185 x 100 x 0.90 = 74.0 phr

For a 100 kg resin batch the charge is 74.0 kg of THPA. Note what the number is not: it is not 74% of a 174 kg batch, and it is not 74 kg of a 100 kg finished mix. Both mistakes are common and both produce a badly under-cured part.

Example 2: resin diluted with a reactive diluent

This is where most formulations go wrong. Adding a low-viscosity epoxy diluent changes the epoxy equivalent weight of the blend, and the anhydride dose has to follow.

Blend: 80 g of resin at EEW 185 plus 20 g of a reactive diluent at EEW 130.

Epoxy equivalents in 100 g = 80/185 + 20/130 = 0.4324 + 0.1538 = 0.5863 eq

Blend EEW = 100 / 0.5863 = 170.6 g/eq

phr = 152.15 / 170.6 x 100 x 0.90 = 80.3 phr of THPA per 100 parts of blend

The dose rose from 74.0 to 80.3 phr, an increase of about 8.5%, because the diluent brought more epoxy groups per gram than the resin it displaced. A formulator who adds a diluent to cut viscosity and leaves the anhydride charge untouched has changed r downwards without noticing.

Example 3: a blended anhydride

When two anhydrides are blended, the blend AEW is the weighted average by mass:

AEW(blend) = 100 / (w1/AEW1 + w2/AEW2)

For a 50:50 blend of THPA (152.15) and HHPA (154.16):

AEW = 100 / (50/152.15 + 50/154.16) = 100 / (0.32863 + 0.32433) = 153.15

The difference from either component is under 1%, which is why cross-family blending does not disturb the dose much, and why a substitution can pass unnoticed on the batch record while changing the cured properties.

Accelerators

Unaccelerated anhydride cure is slow enough to be impractical in most production settings, so an accelerator is added at a fraction of a percent to a few percent on resin.

Accelerator family Typical level Behaviour
Tertiary amines (benzyldimethylamine, DMP-30 type) 0.5 to 2 phr Strong acceleration, works at lower temperature, can darken and can plasticise if overdosed
Substituted imidazoles (2-ethyl-4-methylimidazole type) 0.1 to 1 phr Very strong, long latency at room temperature, sharper exotherm
Quaternary ammonium salts 0.2 to 1 phr Latent at ambient, good for one-pack systems, less colour than amines
Metal salts and chelates Varies Used where a specific latency profile is needed; check compatibility with the filler system

Two practical rules. First, the accelerator changes the gel time far more than it changes the final properties, so dose it for the process window rather than for a property target. Second, an overdosed tertiary amine leaves a plasticising residue that lowers the glass transition and can increase water uptake, which is a difficult defect to diagnose later.

The Cure Profile

A conventional casting cycle for a THPA or HHPA system runs a staged ramp with a post-cure:

Stage Typical condition Purpose
Gel 120 to 140 deg C, 2 to 4 h Vitrification and the start of network build
Cure 140 to 160 deg C, 2 to 6 h Main conversion
Post-cure 160 to 180 deg C, 2 to 4 h Final conversion and glass transition development

The post-cure is the stage most often skipped and the one that decides whether the part passes a heat deflection or a glass transition test. A part that gels and cures but is not post-cured looks and feels finished and is measurably underperforming.

The Errors That Actually Happen

  1. Treating phr as a percentage. 74 phr is 74 parts per 100 parts of resin, which is 42.5% of the total mix, not 74%. This is the single most common arithmetic error in small plants.
  2. Forgetting r. Using the stoichiometric 1.0 when the formulation was developed at 0.90 leaves 10% excess anhydride, which absorbs moisture, raises free acid in the cured part and degrades dielectric performance over time.
  3. Using the nominal EEW instead of the batch certificate. Standard DGEBA is quoted at 185, and real lots can sit anywhere in a band several units wide. On a large casting the difference is small but measurable; on a qualified product it is the variable that explains the drift.
  4. Adding filler or diluent without recalculating. Fillers dilute the epoxy concentration per unit mass and diluents change the equivalent weight. Both shift the anhydride dose, and both are frequently ignored when a formulation is adjusted for cost or viscosity.
  5. Dosing the anhydride on total batch mass. The charge is per hundred parts of resin, and the resin is not the whole mix.
  6. Not accounting for lost anhydride. A lot at 0.5% free acid has 0.5% of its groups already consumed by water. On a 74 phr charge that is roughly 0.4 phr of missing reactivity, which is within the tolerance of most processes but worth knowing about.

Checking a Cure After the Fact

When a casting underperforms, three quick checks tell you whether the dose was the problem.

  • Glass transition by DSC or DMA. A Tg well below the qualification value usually means under-cure or a low r, not a bad resin.
  • Surface tack and free anhydride. A fully cured anhydride system is dry and hard; a tacky surface on a post-cured part points to insufficient anhydride or insufficient accelerator.
  • Batch record arithmetic. Recompute phr from the certificates actually used, including any diluent, and compare with what was charged. Most failures are found in this step.

FAQ

What r should a new formulation start from?

0.85 to 0.90 for a casting system with a tertiary amine or imidazole accelerator. Run a ladder of three levels around it, measure Tg and mechanical properties, and pick the level that meets the property target with the widest process window rather than the highest single number.

Does the anhydride dose change with the filler loading?

Yes, if the resin and anhydride are being dosed on total mix mass. If both are dosed per hundred parts of resin and the filler is added afterwards, the ratio between them does not change. Whichever convention is used, it has to be stated on the batch record.

Can I widen the tolerance by running a higher r?

No. A higher r leaves unreacted anhydride, which is hygroscopic and hydrolyses in service. The system should be dosed to react nearly completely, with the residual anhydride kept below the level at which it affects dielectric properties or exudes.

Is the calculation the same for a liquid anhydride such as MTHPA?

The formula is identical; only the AEW changes, because MTHPA carries a methyl group and has a higher molecular weight: 166.18 theoretical, or 175.7 once the 41.0% anhydride group minimum is allowed for. At EEW 185 and r = 0.90 that puts MTHPA at 85.4 phr against 74.0 phr for THPA, so the number has to come from the certificate rather than from the family name. Product data is on the MTHPA product page.

How does this relate to the epoxy equivalent weight calculation?

They are two halves of the same balance. The epoxy side determines how many reactive groups the resin brings; the anhydride side determines how many the hardener brings. The companion article on epoxy equivalent weight covers the resin side, including blends and diluents in more depth.

Related Reading

Disclaimer: this article gives general formulation arithmetic. Cure schedules, ratios and accelerator levels must be validated by the user in a laboratory and on the production part; nothing here is a recommendation for a specific application.

Contact PolyblueChem for anhydride certificates with the free acid and colour data needed to run this calculation on a real lot.

Scroll to Top