Most accounts of Plasma-Activated Water (PAW) describe its chemistry in general terms. A paper by F. Judée, S. Simon, C. Bailly and T. Dufour of Sorbonne-affiliated laboratories in Paris, published in Water Research 133 (2018) 47-59, instead takes ordinary Paris tap water, treats it for up to 30 minutes, and reports a concentration for each of 16 long-lived species along the way. It then waters lentil seeds with the result. This article follows the pre-print, which carries a notice that its text may differ in places from the published version.
What was the apparatus?
The discharge was a dielectric barrier discharge, one of the families described on Non-Thermal Plasma (Cold Atmospheric Plasma). The paper describes an electrode made of a metal plate carrying a 4*4 matrix of metal rods, each sheathed in a dielectric tube, held 4 mm above the water. It was driven at an amplitude of 12.0 kV AC and 500 Hz. Below it sat 50 mL of tap water in a grounded Petri dish 136 mm across, in a layer 3.44 mm deep.
The rig could run in open air or inside a small hermetic enclosure intended to trap the gas-phase products and force them into the water. The authors' discussion of the chemistry treats the enclosure as the operating case, and they attribute two of their odder curves to it.
How were sixteen species counted?
The count is partly a matter of bookkeeping. By the paper's tally, 11 species came from colour-forming assays (Nessler reagent for ammonia and ammonium, titanium oxysulfate for hydrogen peroxide, the ascorbic acid method for orthophosphates, Griess reagent for nitrite, with nitrate read by an ion probe), 3 from acid titration of the carbonate family, and 2, hydroxide and hydronium, from pH and temperature. Many of the 16 are acid and base forms of the same substance, split by calculation from the measured total using the pH and equilibrium constants.
What changed in the water?
The reported figures, from untreated water to 30 minutes of treatment:
- Nitrate rose from 0.86 ± 0.09 mmol/L to 3.55 ± 0.42 mmol/L, nearly linearly.
- Nitrite climbed from 7.16 ± 0.23 µmol/L to 175.36 ± 9.67 µmol/L at 10 minutes, held level to 15 minutes, then fell to 125.79 ± 6.85 µmol/L.
- Hydrogen peroxide reached 1.85 ± 0.16 mmol/L, following a quadratic rather than a linear curve.
- Ammonium rose from 9.88 ± 3.40 µmol/L to 70.25 ± 0.75 µmol/L; ammonia from 0.42 ± 0.25 µmol/L to 2.12 ± 0.31 µmol/L.
- Bicarbonate fell from 4.13 ± 0.09 mmol/L to 0.98 ± 0.25 mmol/L, most of the loss coming in the first minutes.
- Orthophosphate stayed flat at about 10.84 ± 0.97 µmol/L.
- Temperature rose by 9.03 ± 0.78°C from a 27°C start.
- pH barely moved, from 7.8 to 7.6.
That last line matters for the dossier. The general description of plasma-treated water has dissolved nitrogen oxides making it acidic. In this tap water the authors report the opposite of a dramatic shift, and conclude that the treatment has "no impact on pH but a major one on electrical conductivity." The paper does not test whether the same holds for other waters.
How do the authors explain the curves?
For nitrite, the paper proposes a competition inside the sealed enclosure: oxygen is used up and ozone accumulates, so less nitric oxide forms over time while ozone converts dissolved nitrite into nitrate. For hydrogen peroxide, it points to humidity: relative humidity in the enclosure climbed from 46.63 ± 0.80% to 89.33 ± 0.58% over 30 minutes, and more water vapour means more hydroxyl radicals to pair up into peroxide. The humidity was measured in the enclosure. The oxygen depletion, the ozone build-up and the falling nitric oxide formation, by contrast, are inferred from the liquid chemistry and cited literature rather than measured.
The authors admit they cannot yet explain the bicarbonate loss. Their conjecture is that local heating drives it off as carbon dioxide. They support this with a correlation of r = –0.985 between water temperature and bicarbonate concentration, and note that measuring carbon dioxide in the chamber would be the way to confirm it. That measurement is not reported.
What is the conductivity check?
Conductivity dipped from 647.33 ± 15.07 µS/cm to 614 ± 10.60 µS/cm in the first 5 minutes, then rose linearly to 731.33 ± 19.37 µS/cm at 30 minutes. To see whether their chemistry accounts for this, the authors summed the individual contributions of 18 ions using Kohlrausch's law, taking some concentrations from their own assays and the rest (the mineral salts) from the regional health agency's published analysis of Paris water.
Model and meter agreed with a correlation factor of r = 0.98. In the model, untreated water's conductivity is mostly calcium (41.98%) and bicarbonate (27.96%); by 30 minutes bicarbonate's share has dropped to 6.07% and nitrate's has grown from 9.37% to 35.16%. The early dip is thus read as bicarbonate leaving, the later rise as nitrate arriving.
The authors take the fit as showing that their toolkit is complete for long-lived species. A conductivity sum can only vouch for charged species, however; it says nothing about a neutral molecule such as hydrogen peroxide, and part of its input is municipal data rather than measurement.
What happened to the lentils?
One hundred coral lentil seeds per condition were sprouted in glass jars without soil, soaked and then irrigated twice daily with either untreated tap water or water treated for 15 minutes. Germination was unaffected: 97% against 99%, with half the seeds germinated at 11h16 and 11h29 respectively.
Growth differed. At day 3 the paper gives mean seedling lengths of 1.4 ± 0.7 cm for untreated water and 1.97 ± 0.81 cm for treated; at day 6, 1.87 ± 0.95 cm against 4.28 ± 2.15 cm, an increase it states as 128.41%. The day-3 gain is given as 38.0% in the results and 34.0% in the abstract. The same group's earlier helium plasma jet work is cited as giving 77% after 6 days.
What does this not show?
The agronomic result is labelled a proof of concept by the authors, and the description fits. It covers one species, one treatment time, six days, and no soil. The seeds were already fast and near-complete germinators, so a germination effect could not be seen. The spread is wide: the treated group's day-6 standard deviation is half its mean, and the paper reports that this group's length distribution was too irregular for a Gaussian fit. No statistical test of the difference is reported in the text, nor any repeat of the trial.
Nor is the cause identified. The authors argue from earlier plant literature that nitrite, nitrate, ammonium and hydrogen peroxide (0.7 mM in the 15-minute water) are each at plausible growth-promoting levels, and that losing bicarbonate may help too. But their preliminary tests, adding single ingredients to tap water, found that each helped on its own (hydrogen peroxide +88%, nitrite +62%), and so did merely warming the water (+72%). They state that disentangling the combinations was beyond the paper. Those preliminary tests are mentioned without data or methods.
Connections
The measured species are the long-lived end of the family described on Reactive Oxygen and Nitrogen Species (RONS), and the lentil trial is a concrete instance of the irrigation use listed on Plasma Agriculture. The paper measures water immediately after treatment and does not follow its decay in storage; for that, and for a caution about antibacterial claims, see The Neutralization Problem: Why Plasma-Activated Water Kill Rates May Be Overstated.
Source notes & attribution
- https://arxiv.org/pdf/1802.06886
- https://doi.org/10.1016/j.watres.2017.12.035