To assess what sort of filtration you might want for reducing covid risk, it would be useful to know how the different sizes of aerosol particles contribute to covid spread. Unfortunately, we don't know that much here.

We know that sars-cov-2 itself is ~0.1µm, but it's probably not transmitted as bare virus (Azimi and Stephens 2013):

Some researchers have assumed that the individual virus or bacteria particles are aerosolized and exist suspended as individual organisms; however, it is likely more appropriate to consider the particles as larger expelled droplets that contain aggregates of the smaller infectious particles within.

We also know that there is significant transmission from droplets well under 100µm, since larger ones generally quickly settle out of the air and there's lots of evidence of sars-cov-2 lingering in poorly ventilated spaces.

If you search [what size are sars-cov-2 particles], Google will give you Lee 2020, but I don't think it's useful:

In the case of maximum viral-loading derived from experimental data of COVID-19 patients, 8.97 x 10^-5% of a respiratory fluid particle from a COVID-19 patient is occupied by SARS-CoV-2. Hence, the minimum size of a respiratory particle that can contain SARS-CoV-2 is calculated to be approximately 9.3 µm. The minimum size of the particles can decrease due to the evaporation of water on the particle surfaces.

Stepping through their calculation, first they looked at what fraction of an infected person's saliva is sars-cov-2: typically 3e-7% but a maximum of 9e-5%. Then they ask how big a particle would have to be to contain at least one sars-cov-2 viron, and get 65µm for 3e-7% and 9.3µm for 9e-5%. If you generate a respiratory particle of size 1µm then yes, that's not big enough to contain both sars-cov-2 and everything else in your spit in represenative proportions, and they conclude that this means particles of that size can't contain sars-cov-2. Instead, all this tells us is that only ~0.1% of those particles would contain sars-cov-2.

They acknowledge this ("viruses can be distributed nonhomogeneously inside the respiratory fluid particles; therefore, the minimum size can vary owing to the clustering of viruses in fluid particles") but they don't take it seriously enough to conclude that their whole estimation process doesn't work.

Much better would be to measure airborne particles, and determine what fraction of the sars-cov-2 present in the air is contained in particles of different sizes. Stern et al. 2021 did this with measurements in Kuwait's Jaber Hospital. They used a custom-built device capable of measuring whether sars-cov-2 was present in particles over 10µm, 2.5-10µm, and under 2.5µm. They took 150 samples at the hospital, and 137 didn't detect any sars-cov-2. Of the 13 that did, the size distribution was:

size	number of positive samples
<2.5µm	3
2.5-10µm	7
>10µm	3

While the study also measured sars-cov-2 concentration, the sample sizes are so small that I think it's clearer just to stick with counts.

There has been an enormous amount of research on covid, and I'm not that familiar with the area, so it's possible there are other studies haven't found which conducted size-fractioned sars-cov-2 sampling.

For more data we could look at analogues from other viruses. Azimi and Stephens, 2013 summarize what we know for flu:

On average across all of these studies, we estimate that approximately 20% of influenza virus content is associated with particles in the 0.3-1 µm size range in these recent studies; 29% is associated with the 1-3 µm size range, and 51% is associated with the 3-10 µm size range.

This is calculated by sampling air in places where there are flu patients (hospitals etc), looking at particles of varying sizes, and measuring how much of the total flu virus content of the air was accounted for by each particle size range.

We could assume that the distribution measured for flu is approximately the same as what we would see for sars-cov-2, but as a non-expert it isn't clear to me whether that's a reasonable assumption. Azimi et al, 2020 apply this reasoning to measles, however, which should apply equally to sars-cov-2:

In this study, because of the lack of a reliable source, we assumed that the size distribution of measles viruses in indoor bio-aerosols is similar to influenza viruses. This assumption is based on the fact that both diseases are airborne viral respiratory infections with similar virus sizes ranging between 80-120nm and 100-200nm for influenza and measles viruse, respectively.

Because the sample size for Stern (2021) is so small, and it's roughly consistent with what Azimi and Stephens (2013) see for flu, I'm just going to stick with the flu estimates:

Size	Fraction of airborne virus
0.3-1µm	20%
1-3µm	29%
3-10µm	51%

One interesting factor here is that the size of particles it is most important to remove probably depends a lot on how much outside air you are bringing in. Larger particles settle out of the air more quickly: ~50µm in seconds, ~10µm in minutes, ~2µm in tens of minutes, and ~0.5µm in hours. This means that the fewer air changes you have per hour the larger a proportion of airborne sars-cov-2 will be in very fine particles, the hardest to filter. In other words, the more you are relying on filtering relative to bringing in fresh air, the more important it is that your filter perform well with <2µm particles.