The UV dose d is a measurement of the energy Φe per unit area A that is incident on a surface. The UV dose is the product of the average intensity acting on a microoranism from all directions and the exposure time t:
The UV dose required for inactivation of virtually all pathogens is 40mJ/cm2.
Calculating the velocity of the air flow through the disinfection chamber is difficult because we are dealing with turbulent flow, mainly caused by the ventilation slots and ribs at the entrance and exit of the chamber. Therefore we have to make some assumptions.
The volume of gas inhaled or exhaled from a person's lungs is around 0.5L. Let us assume that we need 2 seconds per breath, then this results in a volume flow Q of 0.25L/s or 0.00025m3/s. If we also assume that it is only laminar flow through a round tube with flow cross-section A○ and inner diamter di, the speed v of the breathing gas flowing through the disinfection chamber is given by:
With the inner diameter of the disinfection chamber of 40 mm this results in a speed of 0.20m/s. The disinfection chamber of the prototype has a length of 20mm. If we assume that the microoranism takes the straight path through the chamber, it needs 0.1s. Together with a surface area of 2x4cm this results in a UV dose of only 0.19mJ/cm2. This means the UVC LED must have a radiant power of 3.2W to reliably kill all pathogens in the chamber.
UVC LEDs with a radiant flux of up to 100mW are currently available on the market. But these high-power LEDs generate a considerable amount of heat and can only be operated with an appropriate heat sink.
The t/A ratio cannot be improved significantly. The volume flow Q is to be regarded as constant. The exposure time t depends on the area A, since the area has an effect on the flow velocity. On the other hand this paper states "Literature supports UVGI exposures of 1 J/cm2 are capable of decontaminating influenza virus on N95 FFRs and exposures as low as of 2-5 mJ/cm2 are capable inactivating coronaviruses on surface". Using a LED with a radiant flux of 100mW we would get a dose of 1,266 mJ/cm2 in the respiratory airflow at assumed flow velocity. So we would have to halve or even better divide the flow velocity by three, which I think is feasible, to deactivate at least corona viruses. And a surface is different to an air stream, as viruses can "hide" in the pores and microscopic irregularities due to their small size. In the air stream, however, they are also enveloped in aerosols, but this is an advantage because these tiny droplets act like microscopic lenses.
Another concern was the formation of ozone but I can rule this out now. Only UV radiation below 200 nm splits oxygen molecules into atomic oxygen. The recombination of the atomic oxygen would then lead to the formation of ozone.