How to Prepare a Catalyst Ink for RDE: Mixing, Sonication and Drying

Water, alcohol, catalyst and a few drops of Nafion. Sonicate, pipette onto the disk and leave to dry. Preparing an ink for a rotating disk electrode (RDE) takes just a few lines in an experimental section.

But compare papers, and those lines start to change. One adds more alcohol; another leaves out Nafion. Some dry the droplet while rotating the electrode; others keep it still. How much does any of this matter? Is there a better way to do it?

As with figure-of-eight and circular motions when polishing glassy carbon, it helps to separate customary practice from what has been compared experimentally. The studies discussed here do not offer a recipe that works for every catalyst. They do help explain what each step aims to achieve, and why copying an isolated number can give a different result.

Our starting point is platinum on carbon, Pt/C, for the oxygen reduction reaction (ORR), for which several studies specifically address preparation. We will also look at oxide catalysts for the oxygen evolution reaction (OER), to examine how far conclusions drawn from Pt/C can extend.

Water and alcohol: what ratio should the ink contain?

The solvent must allow the powder to disperse and be deposited onto the disk. It must then evaporate, leaving a suitable film. Choosing the solvent mixture and choosing the drying method are therefore related decisions.

Pine Research's practical guide [1] suggests water where possible and 20–40% isopropanol for hydrophobic catalysts. It recommends keeping the alcohol content low because alcohol can make deposition more difficult. In particular, on an RRDE with a PTFE gap between the disk and ring, adding alcohol makes the droplet more likely to wet that gap as well, making it harder to coat only the disk.

Experiments with Pt/C help explain why that range alone is not enough. In a comparison of three catalysts [2], two showed a maximum in electrochemical surface area at around 35% isopropanol, while the third showed a less pronounced dependence. The same mixture did not improve every material equally.

Another study of ink formulation and drying [3] selected approximately 24% isopropanol. By contrast, a more recent study of Pt/C and PtCu/C [4] found the most uniform Pt/C films with water:isopropanol ratios of 1:1 and 1:3; for PtCu/C, the greatest uniformity was obtained at 1:3. This comparison used an I/C ratio of 0.3 and drying at 700 rpm. The authors selected 1:3 for further work.

These numbers are not votes for a particular ratio. The catalyst, support, ionomer, drying method and definition of a good result all vary: dispersion, electrochemical surface area or visual uniformity. The shared conclusion is more useful than choosing a percentage: the water/alcohol ratio is part of the preparation method, and its adjustment depends on the material and drying conditions.

Nafion: how much is needed, and what does it do?

In many inks, Nafion is added to help disperse the catalyst and bind it to the disk [1, 5]. But it is not an electrochemically invisible ingredient.

The 2018 report from Toyota Central R&D Labs [5] brings together results from Shinozaki's team on this issue. When they coated Pt/C films with Nafion, they observed a decrease in ORR activity whose magnitude depended on the carbon support. The authors relate this to ionomer coverage of the catalyst; not all Pt/C catalysts responded in the same way.

There are also Nafion-free TF-RDE protocols [6], with specific formulations and drying conditions. In Shinozaki's 2015 study, ionomer-free films with loadings up to approximately 18 µgPt/cm² adhered sufficiently for ORR measurements at room temperature, with no flaking observed [3]. Nafion therefore cannot be considered essential in every case. Nor can removing it be expected to improve every ink: poor dispersion or adhesion creates another problem.

One frequently cited comparison needs care. In Shinozaki's 2015 study, Nafion-free films dried in an isopropanol atmosphere gave a mean specific activity 2.8 times that of the method using Nafion and stationary drying in air for the high-surface-area carbon support, and 2.0 times for Vulcan. However, the loading, deposited volume and drying method also changed. These activity enhancement factors compare two complete procedures, rather than the isolated effect of removing Nafion [3].

Oxides introduce other trade-offs. In a study of Ir and IrO₂ [7], images of deposits on gold foil prepared for a flow cell showed IrO₂ agglomerates without Nafion. At the highest ionomer proportion tested, the authors describe a continuous Nafion layer enveloping most of the agglomerates. This illustrates how the deposit changes in that configuration; the observation was not made on the RDE films.

This literature does not yield a universal amount. It does highlight a precaution when reading recipes: “5% Nafion” may describe the commercial dispersion being added [1]. In Shinozaki and Kokhanov, I/C denotes the ionomer-to-carbon mass ratio [3, 4]; Zhang uses the same abbreviation for the ionomer-to-Pt/C-catalyst ratio [10]. Check which mass is in the denominator before comparing recipes.

Sonication: five minutes, half an hour or longer?

Sonication aims to disperse the ink, but time alone does not fully describe the treatment. A bath and a probe do not deliver the same energy to the sample; power and temperature also matter.

In the study comparing different isopropanol proportions, sonication for less than five minutes gave poorly reproducible results; the authors considered durations above roughly ten to fifteen minutes sufficient for the catalysts studied. They report that extending sonication from fifteen minutes to three hours neither reduced the electrochemical surface area nor caused particle growth detectable by TEM. Their reference recipe included an ice bath; the passage describing the extended sonication test does not specify its thermal conditions [2].

This does not make three hours harmless for every ink. A comparison of bath sonication, probe sonication and mixing [8] found a lower electrochemical surface area with the probe at the highest power tested. Another study observed a loss of area when sonicating without cooling [3].

The results make more sense when their context is retained: “sonicate for fifteen minutes” is an incomplete description without the equipment and thermal conditions. A change in electrochemical surface area also does not, by itself, identify what happened to the particles. Mechanisms proposed by the authors should not be presented as demonstrated damage pathways without measurements that distinguish between them.

How much ink to deposit: volume and loading are not the same

Two droplets of the same volume can contain different amounts of catalyst. Nominal loading depends on ink concentration, deposited volume and disk area. If loading is reported as Pt mass, the platinum fraction of the powder also matters.

There is another, less obvious difference: the concentration calculated when preparing the vial may not match that of the deposited aliquot. A study using X-ray fluorescence (XRF) [9] compared nominal loadings with XRF estimates for inks in different conditions. The authors caution that porosity and roughness complicate absolute estimation, so they interpret their comparisons in relative terms. Their prolonged-storage case involved evaporation and sedimentation; it did not isolate an effect of time. The study illustrates how the estimated deposited mass affects the calculated activity per unit mass.

Should you therefore deposit less to obtain a thinner film? Up to a point. In a series of films made with 20% Pt/C [10], low loadings left parts of the disk uncovered; high loadings produced thicker films and a less accurate assessment of ORR performance. The authors proposed a range for that catalyst and its natural drying conditions. The series did not isolate loading alone: ink concentration and viscosity varied, and keeping the Nafion amount fixed while changing the Pt/C mass also changed the ionomer-to-catalyst ratio. The proposed range cannot simply be transferred to another powder.

Oxides provide a particularly useful example. To compare sixteen materials for alkaline OER, Jung and colleagues [11] chose approximately 0.8 mg of catalyst per cm². They first adjusted the loading with one IrO₂ material until increasing it no longer improved the overpotential at 10 mA/cm². They then applied that loading to the other oxides to keep the comparison consistent.

This shows how they selected the value in that study. It does not establish the optimum for all sixteen oxides, or explain the origin of every recipe using a similar loading.

Loading is therefore a test condition that needs context. As a reporting practice, it is also useful to state the volume and number of droplets: reporting only the final mass leaves part of the procedure undescribed.

Stationary or rotational drying: is there a winner?

This is probably the closest comparison to figure-of-eight versus circular polishing. Drying a droplet seems like a minor detail, until you look at where the catalyst ends up.

In Garsany's 2011 study [12], stationary drying produced Pt/C films with material accumulated at the edge: the familiar coffee ring. Drying while rotating the inverted electrode at 700 rpm produced more uniform films.

For 19.7% Pt/C, they compared seven films of each type. The mean mass activity at 0.90 V vs RHE increased from 0.18 to 0.31 A/mgPt, and the variation between films decreased. The electrochemical surface area changed much less. Preparation had substantially changed the measured activity without changing the starting powder [12].

However, the authors themselves noted that the drying method, including the rotation rate, would require optimisation for other catalysts and loadings [12]. Other studies also show more than one route:

Published procedure	Material and observed result
Rotational drying	For Garsany's Pt/C, better uniformity and less variation than with stationary drying [12]
Slow drying in an IPA atmosphere	Thin, uniform Nafion-free Pt/C films at 4.5 µgPt/cm²; sensitive to operator skill, levelling and humidity [3]
Drying under argon humidified with IPA/water	A documented alternative for Nafion-free Pt/C [6]
Drying without rotation	Selected in Zlatar's iridium RDE protocol [7] because it gave more uniform films under their conditions

Pine's guide also mentions a suggestion from some researchers to dry films slowly in a somewhat humid environment. It presents this as a possibility, not an experimental comparison establishing the best method [1].

Meanwhile, the recent Pt/C and PtCu/C study found coffee rings with water-rich inks while the electrode was rotating at 700 rpm [4]. The authors propose that centrifugal forces and differences in surface tension during evaporation contributed to the accumulation. That is their interpretation of the mechanism; the observation is that rotation did not guarantee a uniform film [4].

These publications do not establish a winner for every case. They support a more specific conclusion: rotational drying is well supported for certain Pt/C inks, but the mixture and drying method work together. A different result with iridium, another substrate and another ink does not refute that comparison.

How do researchers check that the film is suitable?

Published checks combine imaging and electrochemical response. Inspection can reveal accumulation at the edge, uncovered areas or differences between deposits. It does not, by itself, demonstrate that the entire catalyst surface is accessible.

Similar electrochemical surface areas are not sufficient either. In Garsany's comparison, films with relatively similar areas gave different mass activities [12]. This is why it is useful both to inspect the film and to compare curves and independently prepared deposits, rather than only repeat scans on the same disk.

In OER, gas also forms within and around the layer. In a study of Ir/ATO in acidic medium [13], the authors attributed the apparent performance loss primarily to trapped bubbles under the conditions studied. After purging with argon at open circuit, they recovered part of the lost activity. Recovery was incomplete, and the result does not show that every decline in OER performance is reversible.

Taken together, these studies point to a practical conclusion: document the complete procedure, including ink composition and condition, sonication, loading and volume, deposition method, drying and film checks. This is an editorial summary of the variables compared in the sources, not a new protocol validated by ElectroSeek.

A Nafion-free film can work, as can a film dried under rotation or one prepared in a controlled atmosphere. What makes a published recipe useful is knowing which material it worked for, what was compared and how the result was checked.

Materials and equipment for your experiments

On ElectroSeek, you can explore the OrigaTrod RDE, the IPS RRDE system as well as powder catalysts such as IrO₂ or Pt/C for your experiments. If you are looking for equipment or a catalyst, tell us which reaction you are studying, in which electrolyte and with what setup: we can help you compare options and request a quote.

References

[1] Dalton, F. Probing Fuel Cell Electrocatalyst Properties with Rotating Disk and Rotating Ring-Disk Electrodes. Pine Research (2026), §4.2. [2] Takahashi, I. et al. J. Power Sources 195 (2010), 6312–6322. DOI. [3] Shinozaki, K. et al. J. Electrochem. Soc. 162 (2015), F1384–F1396. DOI. [4] Kokhanov, A. A. et al. Catalysts 15 (2025), 1140. DOI. [5] Shinozaki, K. et al. R&D Review of Toyota CRDL 49(4) (2018), 33–46, §§2.2 and 4. Report. [6] Inaba, M. et al. J. Vis. Exp. 133 (2018), e57105. DOI. [7] Zlatar, M. et al. J. Electrochem. Soc. 172 (2025), 106506. DOI. [8] Pollet, B. G. et al. Electrochim. Acta 128 (2014), 292–303. DOI. [9] Chourashiya, M. et al. Anal. Chem. 90 (2018), 14181–14187. DOI. [10] Zhang, Y. et al. Electrochim. Acta 429 (2022), 140953. DOI. [11] Jung, S. et al. J. Mater. Chem. A 4 (2016), 3068–3076. DOI. [12] Garsany, Y. et al. J. Electroanal. Chem. 662 (2011), 396–406. DOI. [13] El-Sayed, H. A. et al. J. Electrochem. Soc. 166 (2019), F458–F464. DOI.