Zirconia Beads or Glass Beads? What Protocols Specify

Zirconia beads, glass beads, steel balls, garnet sand. A published bead beating method almost always names the grinding medium it used, down to a tenth of a millimetre. What it rarely names is how much of that medium ended up in each tube or each well. We read four papers from 2025 and 2026 that describe the beating step in enough detail to repeat, and the pattern is the same in all four: exact on material and diameter, silent on loading.

This article sets out what those four methods actually specify, what that says about choosing between zirconia beads and the alternatives, and how to check the part they leave out on your own bench in about half an hour.

What four recent papers put in the tube

The most precise of the four is a soil DNA method from the Natural Resources Institute Finland. Velmala and colleagues built a lysis matrix from equal parts by volume of three materials: blasting glass spheres of 0.09 to 0.15 mm, ceramic spheres of 0.6 to 0.85 mm, and garnet sand of 0.50 to 1.00 mm. On top of that matrix they added one or three steel balls of 4.5 mm per tube, by hand. The dose scales with the sample: 6 ml of matrix for 2.5 g of soil, 12 ml for 5 g, in 15 ml and 50 ml conical tubes, homogenised three times for 30 seconds at 6 m/s (Velmala S, Tuomivirta T, Pitkänen J-M, Latvala S, Pennanen T, 2026, The Microbe 12:100786, doi.org/10.1016/j.microb.2026.100786).

A fungal DNA protocol for nanopore shotgun metagenomics lists two media side by side. Langsiri and colleagues used glass beads and zirconia beads in the same methods section, with grinding in liquid nitrogen described as a separate route to the same end (Langsiri N, Meyer W, Irinyi L, Worasilchai N, Pombubpa N, Wongsurawat T, Jenjaroenpun P, Luangsa-ard JJ, Chindamporn A, 2025, mSystems 10(6):e01166-24, doi.org/10.1128/msystems.01166-24).

The third paper went further on the beating itself than on the beads. Ge and colleagues compared nine bead beating conditions for animal food microbiomes and resistomes and reported that, among those nine, beating on a Vortex-Genie 2 for 20 minutes performed best. The paper does not state the diameter, the material or the number of beads used (Ge B, McDonald RC, Yang Q, Domesle KJ, Sarria S, Li X, Hsu C-H, Jarvis KG, Tadesse DA, 2025, Applied and Environmental Microbiology 91(2):e02230-24, doi.org/10.1128/aem.02230-24).

The fourth used pre-filled kit tubes, so the beads were not a choice at all. Herbert and colleagues showed that the extraction method you pick changes what comes out of the sequencer, in their case the behaviour of adaptive sampling on nanopore runs (Herbert J, Thompson S, Beckett AH, Robson SC, 2025, Environmental Microbiome 20(1):47, doi.org/10.1186/s40793-025-00704-7).

Zirconia beads rarely work alone

The first thing to say plainly: none of these four papers ran zirconia against glass on the same sample under the same settings. So none of them supports a claim that one material beats the other. Anyone telling you otherwise is not quoting these methods.

What the methods do show is how the hard ceramic fraction gets used in practice. In the soil protocol it is one of three components: fine glass for the small particles, ceramic in the middle range, garnet sand for abrasion, and then a few large steel balls carrying most of the mass. In the fungal protocol, glass and zirconia appear in the same list rather than as competing options. The choice on the bench is usually not zirconia or glass. It is which combination, in what proportion, for the cell wall in front of you.

That is also why the shopping question is narrower than it looks. Our own grinding media page lists glass, zirconia, chrome and stainless steel because labs order them in combinations, not one at a time.

The number nobody reports

Read the four methods sections next to each other and one omission stands out. Only one of the four states a bead count per sample, and that is Velmala’s one or three steel balls. None of the four reports the spread on that number across a run: how many tubes got two instead of one, how many wells came out empty.

That matters because the energy delivered to a sample depends on how much medium is in there with it. Two beads and one bead in neighbouring wells of the same plate are not the same experiment, however identical the instrument settings. A protocol that specifies 0.6 to 0.85 mm ceramic to two decimal places and then leaves the count to a spatula has moved its largest uncontrolled variable out of the methods section and into the hands of whoever loaded the plate that morning.

We looked at how different papers handle this in an earlier article, how many tissue homogenizer beads go per tube or well. The short version: some specify by count, some by mass, and a good number not at all.

A half-hour check on your own plate

You do not need a study to find out where your own loading sits. You need one plate, a balance and about thirty minutes.

1. Load a plate with beads exactly the way you normally do, at your normal pace. Do not slow down for the test, or you will measure your careful self rather than your Tuesday self.
2. Pick twelve wells spread across the plate, not a single row. Corners and centre both.
3. Empty each of the twelve into a weigh boat one at a time. Record the count if the beads are large enough to count, the mass if they are not.
4. Write down the lowest and the highest value, and the mean.
5. Have a colleague load a second plate and repeat. Two operators, same protocol.

Then compare that range against the effect size you are chasing. If your twelve wells vary more between themselves than your two protocol variants differ from each other, the loading step is louder than the thing you are measuring, and no change of bead material will quiet it down.

Fixing it is a geometry problem rather than an attention problem. A mesh plate with an opening cut for your bead size holds one bead per position and releases them together, which is what our 96-well bead dispenser and 384-well bead dispenser do in about eight seconds per plate, for bead sizes from 0.8 to 6.0 mm.

Where none of this helps

Three honest limits. If your kit ships pre-filled bead tubes, as in the Herbert protocol, you do not choose the medium and you do not load it; the consistency is the supplier’s problem, not yours. If your protocol uses no beads at all, the question does not arise: Dar and colleagues ground fungal mycelium with a mortar and pestle and no grinding media whatsoever (Dar GJ, Nazir R, Wani SA, Farooq S, Aziz T, Albekairi TH, 2025, Open Life Sciences 20(1):20221006, doi.org/10.1515/biol-2022-1006). And a dispenser does not choose your material. It makes the count repeatable once you have chosen; the choice between zirconia, glass, steel and a mixture still comes from your sample and your own pilot.

Sources

• Velmala S, Tuomivirta T, Pitkänen J-M, Latvala S, Pennanen T. Scalable isolation of soil genomic DNA from microbial eukaryotes to multicellular micro- and mesofauna. The Microbe. 2026;12:100786. https://doi.org/10.1016/j.microb.2026.100786
• Langsiri N, Meyer W, Irinyi L, Worasilchai N, Pombubpa N, Wongsurawat T, Jenjaroenpun P, Luangsa-ard JJ, Chindamporn A. Optimizing fungal DNA extraction and purification for Oxford Nanopore untargeted shotgun metagenomic sequencing from simulated hemoculture specimens. mSystems. 2025;10(6):e01166-24. https://doi.org/10.1128/msystems.01166-24
• Ge B, McDonald RC, Yang Q, Domesle KJ, Sarria S, Li X, Hsu C-H, Jarvis KG, Tadesse DA. Exploring animal food microbiomes and resistomes via 16S rRNA gene amplicon sequencing and shotgun metagenomics. Applied and Environmental Microbiology. 2025;91(2):e02230-24. https://doi.org/10.1128/aem.02230-24
• Herbert J, Thompson S, Beckett AH, Robson SC. Impact of microbiological molecular methodologies on adaptive sampling using nanopore sequencing in metagenomic studies. Environmental Microbiome. 2025;20(1):47. https://doi.org/10.1186/s40793-025-00704-7
• Dar GJ, Nazir R, Wani SA, Farooq S, Aziz T, Albekairi TH. Optimizing a modified cetyltrimethylammonium bromide protocol for fungal DNA extraction: insights from multilocus gene amplification. Open Life Sciences. 2025;20(1):20221006. https://doi.org/10.1515/biol-2022-1006