A solvent is not a background: choosing between DPG, ethanol, IPM and TEC, and how it reorders what you smell

DPG, IPM and TEC are all recorded with odour strength "none" — so "the solvent smells" is largely the wrong objection. The real effect is on release: ethanol changes mass transfer coefficients, and it does not shift the whole evaporation curve uniformly, it reorders which compounds arrive first. Plus: 763 materials carry an explicit recommended evaluation dilution, spanning 0.01% to 50%.

Diluting a material to 10% in DPG changes both concentration and carrier. The carrier is often treated as blank background, but it changes release rates and can reorder when the components appear.

First, an objection that is half wrong

The objection beginners hear most often is that the solvent has its own smell and will interfere. Half of that is wrong, and the wrong half happens to cover the solvents you actually use.

What our database records for these:

Solvent	Odour strength	Descriptors
Dipropylene glycol (DPG)	none	none recorded
Isopropyl myristate (IPM)	none	none recorded
Triethyl citrate (TEC)	none	none recorded
Diethyl phthalate (DEP)	none	none recorded
Propylene glycol (PG)	none	recorded as odorless
Ethanol	medium	alcoholic; ethereal; medicinal
Benzyl benzoate (BB)	low	sweet; balsamic; oily; herbal; floral; fruity
Benzyl alcohol	medium	floral; rose; phenolic; balsamic; sweet

So the diluents you reach for (DPG, IPM, TEC, DEP) carry no odour record at all. "The solvent will mask the material" does not apply to them. No odour does not mean interchangeable: DPG and IPM sit at opposite ends on all four numbers.

But ethanol does smell (medium strength, alcoholic / ethereal / medicinal), and benzyl benzoate is frankly a fragrance material: sweet, balsamic, oily, herbal, floral, fruity, with a recorded substantivity of 322 hours. Use BB as a neutral solvent and you have quietly added a balsamic fixative to your formula. As for how strong the ethanol itself should be, the database's "1:1.5 in 50% alcohol" is saying something else.

If the standard diluents have no smell, then where is the problem?

The effect is on release, not on odour

It is in how the solvent lets the odorant go.

In 2008 Tsachaki and colleagues published work in the Journal of Agricultural and Food Chemistry using a mechanistic model to calculate mass transfer coefficients for 17 aroma compounds released into a gas-diluted headspace, comparing pure water against a water solution containing 120 mL/L ethanol.

The result: mass transfer coefficients were higher in ethanol/water than in water, both because transfer itself improved and because the partition coefficient shifted. Thermal imaging showed distinctly different convection patterns between the two systems, indicating that ethanol promotes mixing within the liquid phase.

One detail matters a great deal for perfumery. Equilibrium and dynamic conditions can point in opposite directions. A sealed vessel reaching headspace equilibrium is one thing; your smelling strip evaporates continuously into moving air, a dynamic system. The study above measured dynamic conditions, and found that at headspace replenishment rates of one to three times per minute, the presence of ethanol helped maintain volatile concentrations near equilibrium.

This is food science, not perfumery. The system is a water/ethanol solution, not a neat solvent, and the measurement is instrumental headspace, not a smelling strip. The mechanism transfers — a solvent alters release through partition and mass transfer — the numbers do not. We cite it to establish that the effect is real, not to borrow its values.

And it does not shift the whole curve uniformly

If a solvent simply made everything release faster or slower, that would be manageable: adjust the dose. The reality is less convenient.

In 2019 Muñoz-González and colleagues had 10 panelists rinse with rosé wines at three ethanol levels (0.5%, 5% and 10% v/v), each aromatised with six fruity esters (ethyl butanoate, isoamyl acetate, ethyl pentanoate, ethyl hexanoate, ethyl octanoate, ethyl decanoate), and measured intraoral aroma release at 0 and 4 minutes afterwards, publishing in Molecules.

The effect depended on when you looked:

At 0 minutes, straight after rinsing, more ethanol pushed the more volatile compounds out faster, while the less volatile ones actually released less.
By 4 minutes, all six esters released more.

The authors conclude that the extent of the effect depended on the subject, the ethanol concentration, and the type of aroma compound.

For a perfumer, that means the solvent reorders the curve: which components arrive first, which arrive later, which persist. And the reordering varies between people. The article on evaporation curves made the point that everyone's curve differs; here it returns with the solvent as the variable.

So take "the same material in DPG and in ethanol are two different things" literally. Both studies point straight at it.

So what concentration should you use?

This is what beginners most want a number for, and our data holds a rarely-noticed field.

Of the 2,251 materials carrying an odour strength value, 763 (33.9%) also carry an explicit recommended evaluation dilution, phrased as "recommend smelling in an X% solution or less".

Recommended dilution	Materials
≤ 50%	12
≤ 20%	1
≤ 10%	498
≤ 5%	4
≤ 1%	195
≤ 0.1%	44
≤ 0.01%	9

From 0.01% to 50%: a five-thousandfold range.

And it tracks the strength rating closely:

Odour strength	Most common recommendation
Very high	≤ 0.1% or ≤ 0.01%
High	≤ 1%
Medium	≤ 10%

Two limits to state plainly. First, this is a subset: only 2,251 of the 21,494 materials with supply information carry an odour strength value, so "33.9%" is a share of those 2,251 and cannot be generalised to the whole database. Second, the recommendation is a compiler's trade convention, not a measurement — it reflects professional evaluation practice, not a threshold determined by an experiment.

Even so, the direction is clear: 10% is the default, but nearly a third of materials need to be weaker than that. My own first batch of materials all went into one dilution, and the strong ones read as pure harshness until I remade them weaker. Evaluate everything at 10% and the materials that belong at 0.1% will forever read as "harsh, can't pick out any detail". That verdict is an artefact of your concentration choice; the material never had a chance.

Choosing between the four

Solvent	Volatility	Own odour	Suited to
DPG	essentially non-volatile	none	The default for evaluation dilutions. Cheap, odourless, broadly solubilising
Ethanol (95%)	high, it leaves	medium, for the first minutes	Finished products and formula dilutions. If the product is alcoholic, evaluate in ethanol
IPM	essentially non-volatile	none	Oil-based products, anything going on skin
TEC	essentially non-volatile	none	An alternative where DPG will not dissolve something; also used in finished products
Benzyl benzoate	low	low but real (balsamic)	Not a neutral solvent. Treat it as a fixative material

About "essentially non-volatile". DPG, IPM and DEP all carry a recorded substantivity of 400 hours. But 400 is the single most common value in that field (18.5% of all entries) — a conventional top of scale rather than a measurement. The article on evaporation curves already established that this field's distribution is dominated by rounded buckets. Read it as "does not meaningfully evaporate on the timescale of the test", not as "exactly 400 hours".

Five working rules

Use DPG for evaluation dilutions, unless your product is alcoholic. It is odourless and non-volatile, and it keeps quiet while you judge the material.
Whatever solvent the finished product uses, do the final assessment in that solvent. A formula perfectly balanced in DPG will reorder itself in ethanol; that is the direct consequence of the second study above. This step cannot be skipped.
Treat solvent and concentration as required fields in a note. The note format in the article on record keeping already has both fields; this article is the reason for them. Without them, "sandalwood: warm, milky" cannot be compared with any other note.
Keep results in different solvents apart. The same material at 10% in DPG and at 10% in ethanol are two records made under different conditions; put them side by side and they will only mislead you.
Give ethanol solutions time. A freshly made ethanol dilution carries an obvious alcoholic top for the first minutes (the medium strength in the table above), and the balance between components shifts as it stands. Wait at least 24 hours before evaluating; finished formulae usually need considerably longer.

A solvent is not the tray the material is served on. It is the table. Change the table and the same dishes arrive in a different order. In perfumery, the order is the work.

References

M. Tsachaki et al., Effect of ethanol, temperature, and gas flow rate on volatile release from aqueous solutions under dynamic headspace dilution conditions, Journal of Agricultural and Food Chemistry, 56(13), 5308–5315 (2008). PMID 18529063

C. Muñoz-González, M. Pérez-Jiménez, C. Criado & M. Á. Pozo-Bayón, Effects of ethanol concentration on oral aroma release after wine consumption, Molecules, 24(18), 3253 (2019). PMID 31500122

Next up is fixation: what a fixative actually does, why "add a musk and it will last" is usually wrong, and how low-volatility materials reshape the whole curve rather than just its tail.