Battery-grade dimethyl carbonate is the clearest example in the whole solvent business of a product whose value lies not in its purity but in the size of its impurity budget. A lot at 99.95% by GC and a lot at 99.50% differ by 0.45 percentage points of volatiles; what actually separates them in a cell is 180 ppm of water and 400 ppm of methanol. This article works through where those limits come from, how the moisture budget of an electrolyte is spent across its components, how the numbers should be measured and sampled, and what a cell failure caused by out-of-specification solvent looks like on the line.
For the parent product, its manufacturing routes and the three-grade specification, see Dimethyl Carbonate: Properties, Grades and Manufacture. The material is catalogued on the DMC product page.
1. What DMC does in the electrolyte
A lithium-ion electrolyte is a compromise. The salt needs a medium with a high dielectric constant to dissociate; the ions need low viscosity to move; the solvents must be stable against a 4.3 V cathode and a graphite anode; and the mixture has to stay liquid from -20 degC to 60 degC. No single solvent does all four, so the industry blends a cyclic carbonate with linear ones.
| Solvent | Melting point (degC) | Boiling point (degC) | Dielectric constant | Viscosity (mPa s) | Role |
|---|---|---|---|---|---|
| Ethylene carbonate | 36 to 38 | 248 | about 90 | about 1.9 at 40 degC | dissociation; SEI formation |
| Propylene carbonate | -49 | 242 | about 65 | about 2.5 at 25 degC | dissociation, wider liquid range than EC |
| Dimethyl carbonate | 2 to 4 | 90 | about 3.1 | about 0.59 at 25 degC | low viscosity, low melting point |
| Ethyl methyl carbonate | -55 | 107 | about 2.9 | about 0.65 at 25 degC | low viscosity, low volatility |
| Diethyl carbonate | -43 | 126 | about 2.8 | about 0.75 at 25 degC | lowest volatility of the linear set |
Values are typical literature figures for orientation, not specification points. The pattern to read is the trade-off: the cyclic carbonates have high dielectric constants and are either viscous (PC) or solid at room temperature (EC); the linear carbonates are fluid and stay liquid well below zero but barely dissociate anything. DMC is the cheapest and lowest-viscosity member of the linear set, which is why it typically occupies 30 to 50% of the solvent package in cells for consumer electronics and in blends where cost matters.
2. Where the 20 ppm water limit comes from
It comes from the salt. Lithium hexafluorophosphate reacts with water:
LiPF6 + H2O -> LiF + POF3 + 2 HF
The stoichiometry is one to one on a molar basis, and the arithmetic is unforgiving:
| Quantity | Calculation | Result |
|---|---|---|
| Water in 1 kg of electrolyte at 20 ppm | 20 mg | 1.11 mmol |
| LiPF6 consumed | 1.11 mmol x 151.91 g/mol | about 169 mg |
| HF released | 1.11 mmol x 2 x 20.01 g/mol | about 44 mg |
| Water in 1 kg at 200 ppm | 200 mg | about 1.69 g of LiPF6 consumed, about 0.44 g HF |
That is the whole argument. Forty-four milligrams of hydrogen fluoride in a kilogram of electrolyte is enough to attack the cathode surface, dissolve transition metals into the electrolyte and shift the SEI composition; at the technical-grade limit the numbers are ten times worse. And the water that matters is not only the water added as a component – it is the sum of the moisture carried by every input and every surface.
3. The moisture budget of a formulated electrolyte
A practical budget, with the solvent at 40% of the package:
| Input | Typical contribution | Comment |
|---|---|---|
| DMC at 20 ppm water, 40% of blend | 8 ppm | battery grade is specified partly to make this term small |
| Ethylene carbonate, dried | 5 to 10 ppm | solid handling; melts and picks up moisture |
| Other linear carbonates | 5 to 10 ppm | same specification logic |
| LiPF6 as supplied | 10 to 20 ppm | salt is the largest single term in many plants |
| Additives (VC, FEC, PS, LiDFOB) | 1 to 5 ppm | small volume, but often stored poorly |
| Headspace, lines, filtration | variable | the term that ruins the budget after a maintenance stop |
Two consequences follow. First, buying DMC at 20 ppm rather than 200 ppm buys back most of the moisture headroom, but it does not give licence to be careless elsewhere – the salt usually contributes more than the solvent. Second, the last row is the one that fails audits: an electrolyte plant that meets its budget for months will blow it after a filter change or a line break, and the water result will be blamed on the solvent supplier because the solvent is the component with a 20 ppm line on the certificate.
For the additives side, see our articles on fluoroethylene carbonate and lithium difluoro(oxalato)borate, both of which have their own moisture sensitivity.
4. Methanol: the second critical impurity
Methanol is the cheapest possible contaminant of DMC, because every commercial route either uses it as a raw material or produces it as a by-product. It is also the one that does the most subtle damage.
- It reacts with lithium. Methanol on the anode surface gives lithium methoxide and hydrogen; the methoxide is not a useful SEI component.
- It shifts the solvent structure. Methanol is a hydrogen-bond donor, and even a few hundred ppm change the local solvation of Li+, which shows up in conductivity at low temperature rather than in any single room-temperature test.
- It is a marker of the drying stage. A lot at 400 ppm methanol against a 100 ppm specification is telling you the final distillation ran short or the drying train was bypassed – which means the water figure on the same certificate deserves a second look.
The 100 / 200 / 500 ppm ladder between battery, high-purity and technical grades is effectively the price of that final stage, which is why grade substitution in a cell plant is never a straightforward cost saving.
5. Colour, and what 5 APHA really measures
Colour is the specification that looks cosmetic and is not. In DMC the colour bodies are unsaturated and peroxide species formed by oxidation during storage or by contact with hot metal surfaces. They are present at levels far below what a GC trace resolves, and they decompose on the cathode at high potential, contributing to gas evolution and to impedance rise.
Practically, APHA 5 means the product was kept clean from the reactor to the drum: passivated stainless steel, no rust, no hot spots in the transfer line, and an oxygen-free headspace. Once formed, colour bodies are not removed by drying, which is why a downgraded lot cannot be re-qualified to battery grade simply by passing it through a molecular sieve.
6. Sampling and analysis: where good specifications go wrong
A 20 ppm water specification is meaningless if the sample is taken in the open. Three practices matter.
| Step | Requirement | Failure mode if ignored |
|---|---|---|
| Sample container | septum-sealed, pre-dried, headspace-free | atmospheric moisture ingress gives 50 to 200 ppm in minutes |
| Sampling point | closed loop, dry nitrogen purge, no plastic tubing | tubing permeation and dead volume |
| Karl Fischer method | coulometric for 20 ppm, oven or direct injection as agreed | volumetric KF lacks resolution at this level |
| Methanol method | headspace GC, calibrated at the ppm level | direct injection distorts the light alcohols |
| Colour | Pt-Co against freshly prepared standards, in a defined path length | visual comparison drifts with the observer |
The pattern in complaints is consistent: the supplier’s retained sample passes and the customer’s drum sample fails, and the difference is almost always the sampling, not the product. Agree the sampling procedure in the quality agreement before the first dispute, not after it.
7. Failure signatures in the cell plant
| Symptom | Likely solvent-related cause | What to check |
|---|---|---|
| Gas evolution during formation | water above budget | KF on all inputs, not just the solvent |
| Capacity fade with rising impedance | HF attack, metal dissolution | water plus salt quality plus cell storage humidity |
| Poor low-temperature performance | methanol or an unbalanced solvent ratio | headspace GC on the blend, conductivity at -20 degC |
| Colour in the blended electrolyte | colour bodies carried in with a solvent | APHA on each incoming solvent, not on the blend only |
| Inconsistent batches from the same recipe | moisture ingress after a maintenance stop | line drying procedure and commissioning checks |
8. Buying battery-grade DMC
- Specify numbers, not names: water 20 ppm max, methanol 100 ppm max, colour 5 APHA max, GC 99.95% min, carbonate 99.99% min.
- Require the packaging to be inert and dry – nitrogen-blanketed internally coated steel drums or ISO tanks, with the blanketing stated on the delivery documents.
- Agree the sampling and analysis method in writing, including who pays for a third-party referee analysis.
- Ask for the lot’s age and for the remaining certification period; battery grade does not improve in a drum.
- Do not re-use opened drums across campaigns; a partially used drum is a moisture source even if it was nitrogen blanketed at filling.
References and scope note
Stoichiometric arithmetic is reproducible from standard atomic weights (water 18.02, LiPF6 151.91, HF 20.01). Physical property ranges are typical literature values for orientation and are not specification points. No claims are made about cell performance; battery design and formation are outside the scope of a solvent article, and nothing here is a substitute for your own qualification testing.
Technical and commercial orientation only. It is not safety, regulatory or engineering advice for any specific plant.
PolyBlueChem supplies battery-grade dimethyl carbonate at 20 ppm water and 100 ppm methanol, with lot-specific certificates and nitrogen-blanketed packaging. Contact polybluechem@163.com for specifications, samples or a quality agreement.