How Much UV-C Does It Take to Disinfect SARS-CoV-2?

May 13, 2020

by Dave Conklin

SARS-CoV-2 is the virus which is causing the Covid-19 pandemic. As far as I know, no one has yet determined, by direct measurement, how much UV-C radiation is required to deactivate the SARS-CoV-2 virus. But two recent publications draw inferences about SARS-CoV-2 sensitivity to UV-C from studies of other viruses.

A March 2020 publication by Kowalski et al., includes a table of the doses required, for each of 8 different Coronaviruses, to inactivate 90% of the pathogen (the D90 dose). The doses range from 0.7 mJ/cm2 to 24.1 mJ/cm2. For reference, the specification for the 2-minute exposure produced by the DayZero UV-Box is “at least 40 mJ/cm2”. Publications are cited for each number in the table, ranging in date from 1978 to 2007. Kowalski himself is the chief scientist at a company, PurpleSun, which manufactures UV disinfection equipment. His two co-authors are medical doctors at Weill Cornell Medicine of Cornell University.

An April 2020 publication by Derraik et al., is a very down-to-earth article about how disposable personal protective equipment may be safely reused, by following a 2-step protocol. The first step is storage for at least 4 days to allow the virus to disintegrate naturally. The second step is treatment by one of a) UV-C, b) heat, or 3) a disinfectant chemical. The paper cites 360 mJ/cm2 as the D90 dose for SARS-CoV-1 from Kowalski’s 2009 Ultraviolet germicidal irradiation handbook UVGI for air and surface disinfection. (Kowalski is the same Kowalski who is the chief scientist at PurpleSun.) Derraik et al. estimate that “the minimum applied UVC dose for effective deactivation of SARS-CoV-2 on N95 FFRs would likely be close to 1,000 mJ/cm2, particularly in light of the mask’s porous surface (as compared to a smooth surface material), as shown by Heimbuch & Harnish’s 2019 study”. FFR stands for filtering facepiece respirator. Derraik et al. recommend a disinfection dose of 2000 mJ/cm2 applied to each side of N95 FFRs, fully 50 times larger than the dose (40 mJ/cm2) commonly deemed adequate for disinfecting water.

Thinking that the high dose recommended by Derraik et al. might be based on a misunderstanding of the size of the D90 dose for SARS-CoV-1, I wrote Kowalski to find out whether the D90 for SARS-CoV-1 is really 360 mJ/cm2 as cited by Derraik et al., or 24.1 mJ/cm2 as in Kowalski’s newer paper. Here is what Kowalski replied.
“I have revised my interpretation of the Darnell results based on an improved understanding of the disinfection process. The high doses in Darnell (using the PBS data) produce a tail at 20 minutes and beyond. Since we are interested in the D90 value his data is problematic. I have modeled his data with a tailed survival curve and extrapolated the D90 value as about 241 J/m2. This is still likely to be a conservative value but I left it at that. See my latest article on The Cluster Model of Ultraviolet Disinfection for more information.”

Derraik et al. cite a 2018 paper by Mills et al. for a dose of 1000 mJ/cm2 as effective for 99.9% disinfection of H1N1 viruses (“swine flu”) on N95 masks. I am not a virologist, but the Wikipedia entry for “H1N1 virus” identifies it as “Influenza A virus subtype H1N1”. It was an H1N1 virus that caused the swine flu pandemic in 2009. In the Kowalski Handbook, Influenza A viruses have D90 values from 17 t0 48 J/m (1.7-4.8 mJ/cm2), lower than the 241 J/m2 now estimated by Kowalski for SARS-CoV-1. That is, the Covid-19 virus is harder to “kill” than the swine flu virus. So, relative to Mills figure for the swine flu virus, Derraik’s recommendation of 2000 mJ/cm2 is plausible. But relative to the commonly used 40 mJ/cm2 for disinfection of drinking water, a 50 times larger number seems high.

The bottom line is that after reading the new Derraik et al. and Kowalski et al publications, and after getting Kowalski’s clarification of the D90 dose for SARS-CoV-1, my guess is that Derraik’s 2000 mJ/cm2 dose will do the job for disinfecting for SARS-CoV-2, but it is possible that somewhat lower doses might be adequate as well. 40 mJ/cm2 may well be too little.

The automatic timer on the DayZero UV-Box products from our second pilot run is set for 2 minutes. That’s more than adequate, according to our measurements, to deliver a UVC dose of at least 40 mJ/cm2 to unshaded surfaces anywhere inside the box. But it is much too short to achieve the 2000 mJ/cm2 dose recommended by Derraik et al. for SARS-CoV-2, or even the 1000 mJ/cm2 dose cited by Derraik et al. in Mills’ work on the swine flu virus.