Is Ambient Far UVC light a safe and effective approach to reducing SARS-CoV-2 transmission?

Summary: Yes, in theory.

Far UVC is the shortest set of wavelengths in the UVC range i.e. 200-220 nm out of 200-280nm. This range is also sometimes incorrectly referred to as Far UV, which is the range from 100 to 200nm that produces ozone. The wavelengths are so short that they do not penetrate skin very deeply. In fact, most of the light is absorbed in the outer layer of dead skin cells that contain only proteins rather than DNA material that can be damaged. Moreover, the wavelengths cannot physically penetrate the teary liquid layer of the eye so there is no way for it to cause damage. These physical theories have been confirmed in various studies showing that longer UVC wavelengths cause carcinogenic damage, whereas Far UVC does not.

Unlike the Reflow Labs Airsafe, which is only useful for the air (save for precipitation of viral material from the air onto surfaces), a safe ambient UV light would be able to quickly and continuously disinfect both air and surfaces in total silence.

There are two types of lighting devices that are used for generating Far UVC: light emitting diodes (LEDs) and excimer lamps.


LEDs have rapidly overtaken incandescent and fluorescent lights for white indoor lighting as their costs have plummeted, allowing companies to take advantage of their high radiant efficiency around 50%. LEDs are also great in their ability to have a narrow spectral output, which is critical in this application. A little too far to the left (below 200 nm), and you start producing toxic ozone. A little to far to the right and you get carcinogenic and eye-damaging mid-range UVC.

Sounds great, but unfortunately, not all LEDs are the same. The LEDs that emit light anywhere in the UVC range are about 50-100x more expensive to produce than 'white' (blue LED + phosphor pump) LEDs. This actually would not be a dealbreaker -- what really gets you is that on top of that the radiant efficiency is incredibly low, around 2%, or 25x worse. This leaves us with costs that are 1,250-2,500 times higher than white LEDs for the same radiant power i.e. on the order of $1-2k for a single watt. The bleeding edge research in this space has demonstrated efficiencies as high 5% and people have worked on improving this for decades with little progress. It would be fantastic if someone figures out how to make it work, but for now it's a non-starter. Be wary of any device with UVC LEDs as they may focus on the power consumption of device without pointing out the radiance is about a fiftieth of that amount. As of today there aren't really any commercially available products with Far UVC LEDs to speak of so this shouldn't be an issue.

Far UVC Excimer Lamps

Excimer lamps are a bit more cost-effective than LEDs, but it seems to only be by a factor around 2-3x. There are two types of excimer lamps that produce Far UVC: krpton-bromine (KrBr), which has an emission peak at 207nm, and krypton-chlorine (KrCl), which has an emission peak at 222nm. Between these 2, it is only the KrCl that are readily available. This could either be due to cost, or the spectral output. Unlike LEDs that can have a single narrow peak, excimer lamps have a primary peak, but may also have a small amount of light across a wider range of wavelengths. This trailing power might be a small fraction of the main peak, but for human exposure, even a small amount of the wrong UV wavelength can pose a large risk. Additionally, 207nm is closer to sub-200nm range where ozone would start being produced.

Far UVC Excimer Lamp Efficacy

The amount of power needed to inactivate coronaviruses is well known for both fluorescent UVC and excimer UVC. The amounts do not vary much between the different wavelengths, nor do they vary much among different enveloped viruses that are all around the same tier on the disinfection heirarchy. While the biological particulars matter a lot to humans, from a physical perspective there is little variation. Looking across a number of papers that have evaluated different strains of coronaviruses and other enveloped viruses, we will take an energy value of 3 mJ/cm^2 as the amount needed to inactivate 99.9% of SARS-CoV-2 in air, and 0.75 mJ/cm^2 as the amount needed to inactivated 90%.

You might wonder how we can arbitrarily select these values when some papers might say 2.1 mJ/cm^2, others might show 3.4 mJ/cm^2, and the reason is that there actually is no set value because the question is not well-defined. Some of the factors not being considered include relative humidity, temperature, aerosol size distribution, and intensity i.e. the time in which the energy is delivered. With perhaps fewer than a dozen papers (there is a lot more research on surfaces -- not because they are more important, but the simple reason that it is a far easier experiment to conduct) we cannot hope to be more precise, nor is there any need to for reducing transmission. We are not considering sterilization. In a real world environment outside of a lab what matters most is the viral load. It is not a binary problem. What we need to figure out is if we are getting closer to the 1-2 range, or 0.01-0.02, which would clearly be inadequate.

We can work backwards from these numbers to determine the power requirements, compare them to existing products, and then evaluate whether there is a safety risk to the trailing power outside of the Far UVC range.

Let's suppose we want 90% inactivation every 5 minutes, which gets us 99.9% in 15-20 minutes. How many watts of Far UVC would we need per 1000 square ft? 1000 square ft is about 930,000 cm^2. If we want each cm^2 to receive 0.75 mJ every 10 minutes we need roughly 700,000 mJ = 700 J total in 5 minutes is only 2.3 watts, and with a typical excimer lamp radiant efficiency around 10% a 23 watt bulb would do the trick. Let's evaluate how the different products stack up and how safe they.

Far UVC Excimer Lamp Vendors and Safety Risks

Larson Electronics offers a wide number of Far UVC excimer bulbs and fixtures. Let's consider their 40 watt bulb with a 10,000+ hour rate lifespan and 360 degree output. The length of the bulb is 5.43 inches, its diameter 1.41 inches, and the stated UV intensity is 40 microwatts per cm^2 at a distance of 1.57 inches. The price at the time of this writing (August 2020) is $1,225. This is great since it lets us confirm whether the theoretical efficiencies of this type of bulbs are achieved, and if not, how much we need to correct.

Let's run the numbers. At a distance of 1.57 inches from the bulb's surface we have an overall diameter of 4.55 inches. If we want the project surface of the light we need to consider that the beam diverges along its length. If we assume a wide 120 degree divergence we have a 60 degree angle off normal on either side. The tangent of this angle times the distance of 1.57 inches gives us an extra 2.72 inches on either side. However, the bulb itself appears to only account for around half the overall fixture's length so we will take the total length of the project light to be 8.15 inches. Together with the overall diameter we get an area of 117 in^2, or 752 cm^2. Multiply by the radiant intensity we arrive at a total radiant power of 30,000 microwatts, or 30 mW, which would be less than 0.1% efficiency, which is absurd. Even LEDs are more than 10x this efficient. Most likely, the vendor made a mistake and meant to write milliwatts rather than microwatts. The distance might have also been from the center of the bulb rather than the surface, and the light might not diverge as much as we guessed, which all together would leave us with an efficiency around 20%, suggesting that efficiency is up.

Are these lights safe? Larson Electronics is kind enough to provide a high resolution spectrum chart allowing us to see for ourselves. They state in the product description that the lights have "no harmful UVC emissions" and "No Block Off/Bandapss/Cutoff Filters) Needed [sic]", but the spectrum chart they provide indicates those claims are false. Throughout the non-far UVC range starting at 230nm all the way up to 260nm, the intensity appears to only be about a twelfth of the 222nm peak. Compared to the triangular section of the peak with a base width around 10nm, which would be a 5nm wide rectangle we have half as much harmful UVC as Far UVC. That seems a bit off just eyeballing the graph so let's call it 80% Far UVC and 20% non-far UVC (sidenote: although any Far UVC lamp that is spec'd appropriately should exceed the safety limits for the Far UVC portion as well, we are not concerned about those since those standards are too conservative and do not take into account the recent research showing that Far UVC is harmless).

How does this 20% high-risk UVC compare to exposure safety standards? Perhaps it's insignificant. If 8 watts of irradiance from the 40 watt bulb hit the ground, we won't be able to easily spread that out uniformly over 1000 square ft like in our theoretical example. With a 120 degree spread and a 12 ft ceiling, a sixth of that power will end up in a 4.2 ft diameter circle (on the ground), and this is the area we care about since someone might end up sitting at a table directly under the light (not to mention that the intensity at eye level will be 4x higher). Our circle on the ground has an area of 14 ft^2, or 13,000 cm^2. Distributing 20% of our 8 million microwatts across this area yield is 122 microwatts per cm^2. Unfortunately, the National Institute for Occupational Safety and Health (NIOSH) advises that power levels for all day exposure do not exceed 0.2 microwatts/cm^2. So the output of harmful UVC is 610 times higher than the recommended exposure limits for skin. That's scary.


Is there a safe alternative? Ushio America also provides Far UVC excimer bulbs, and they have partnered with Acuity Brands to supply these exclusively for general illumination purposes in North America. They have 12w, 20w, and 300w bulbs with pricing of the same order of magnitude as the Larson Electronics bulbs. They mention a proprietary filter for 230nm and above, but they do not provide details on it. Based on our calculations above it would be necessary to have 99.9% attenuation and we cannot comment on the materials science feasibility of achieving such performance while still letting the far UVC through. There could also be some variations within excimer bulbs such that some of them can get away with less filtration. This is the brand we would advise looking at within Far UVC (we have zero affiliation with the various companies mentioned in this article). It seems they might still be working on scaling production since they have stated a planned release for late 2020.

Another brand we looked at was Sterilray. They have a rather interesting array of claims that go alongside their products. They have a product similar to the Larson 12w bulb at a price of $9,250, though with a 20% discount at high volumes (could be somewhat more powerful -- this is not listed on their website. It has a non-mirror-polished reflector, which will cause around 50% losses for over half the light since the bulb is recessed. They had a unique way of describing it: "This unit even on low power will continuosly emit 10 x 15th pathogen killing photons every second. For perspective, that’s about 30,000 times the number of trees on the planet." (presumably, they mean 10*10^15)

Is that a lot? Multiplying the frequency of Far UVC by Planck's constant indicates that a single photon of 222nm light only has 1.42*10^-19 joules so they are claiming an output of 0.0014 watts on low. It seems they made an off by 1,000 error or an off by since 14 watts seems high for a low setting since that would correspond to 70 watts given our 20% estimated efficiency. These things happen. But more importantly, the fact that they provide the ability to adjust the power level, mentioning that the high power is intended for unoccupied spaces should be of concern. Why would you need to lower the power if it's safe?

Sterilray also makes the GermBuster-Air Canister, a similar device to the Reflow Labs Airsafe. Unfortunately, it offers less than 70% as much airflow in a package that's roughly 3 times bulkier, and a price that is 6-8x higher based on the pricing of their bulbs. This might be accepted by hospitals, but it is a non-starter for most businesses. We'd rather outsell them 1000x than finding the most desperate/vulnerable subset of customers and making 6-8x on those sales.

We also sent them another email to find out if they use a filter for the >230nm wavelengths. This was their entire reply:

"The patented Excimer Wave Sterilray wavelength also has a patented power supply so that anything 250W and above only produces the 222nm wavelength. Therefore, no filter is needed. It makes a lot more sense than producing an inefficient bulb which only lasts 3,000 hours; our Excimer Wave lamps are warrantied for 30,000 hours. The inefficient bulbs also do not isolate the safe wavelength and still produces all the dangerous wavelengths hoping that a filter, that will eventually wear out, will catch them all. It doesn’t catch them all and how do you know when you need to replace the filter? Does this help?"

A power supply supplies power so the first sentence is not right or wrong, it's nonsense. They go on to to suggest that an ordinary fluorescent UV bulb lasts 3,000 hours so that their 30,000 hour lifespan seems better, but the actual lifespan of fluorescent UV bulbs is closer to 8,000 hours. They also suggest that ordinary UVC bulbs are used with filters to produce only Far UVC, which is a complete fabrication since there is no such thing as a 254nm bulb with a filter to only emit Far UVC since there is barely any Far UVC to begin with. They compared themselves to a fictional product/technology. Ironically, if safe filters for excimers bulbs do exist, and if 254nm UVC bulbs emitted even a tenth of their power in the Far UVC range, this fictional approach would actually be far better than the excimer bulbs since it would cost over 100x less for just a 10x hit on power.

On their website they make similarly absurd claims such as saying that the excimer bulbs are 10-1000x more effective at eliminating pathogens than typical UVC bulbs -- with viruses the intensities needed vary by a factor of around 2x and go both ways depending on the virus. Finally, they state that the "lamp does not contain mercury or other dangerous materials", except for the fact that a KrCl excimer lamp has chlorine in it, a lethal gas. If you dig into anything you'll probably find something toxic somewhere so it's best to avoid breaking things. The magnetron in your microwave for instance contains berrylium.

We hate to say it since it splashes back on us to be this critical, but these are extraordinary times so we feel obligated to make it explicit and jump on the grenade: Sterilray is a slimeball company through and through, they are a threat to public safety, and we hope to see them stopped, if not prosecuted to the full extent of the law. We have no qualms making these statements from a liability perspective since everything we have written is true.

Are UVA LED Lights a Good Option?

UVA LEDs don't face the efficiency problem of UVC LEDs and they are also much safer for human exposure. However, they have practically zero effects on viruses. We spoke with an executive in the restaurant industry who had spoken with one of these companies and it seems they have been insinuating that reducing bacterial levels in the air somehow had a 2nd order effect on viruses. It doesn't. A good trick for making sure you aren't being sold something flagrantly false is communicating via email where there is a written record of the statements made. On the phone people may be tempted to wax lyrical. Another thing to note is UVA LEDs are cheap so it's not clear why this company is price-gouging and selling at ~$8-12k per location costs for eateries. Finally, if you've ever heard someone concerned about the effects of harsh blue light from staring into a computer monitor all day, this long wavelength UVA is just like that, except slightly worse for you. You won't get cancer or go blind, but some people might find it uncomfortable and some people will be concerned if it may slightly accelerate the long-term degradation of vision that comes with age. UVA is not for viruses.

In Review

There is reason to be optimistic and there are reasons to be concerned about Far UVC. On one hand, if proper filtration above 230nm exists and is used, leaving aside cost, we would consider this a superior technology to the Reflow Labs Airsafe. We do not agree with the existing safety standards for Far UVC and expect these will be updated to take into account the data showing they are safe -- it is not just that data exists showing lack of harm from Far UVC, but the physics of the light itself prevent physical entry deep into the eye or deep under the skin. We hope it will be possible to educate the public on the safety of these lights so that they can be widely used. The costs are higher than using the Reflow Labs Airsafe, but they offer the advantage of hitting exposed surfaces and operating in total silence.

That being said, there is one major risk to companies that might consider ambient UV lighting -- how will your customers know that the particular light in place is one of the safe ones that has filtered the mid-range UVC? You could put up a poster of this or that certification, but at this rate, it seems no regulatory body is stopping the companies (whether there is a risk of criminal negligence or just civil charges we do not know) selling these lights. This risk is why we have no problem and in fact consider it our duty to call out companies that may be selling dangerous products -- they make it impossible to trust the ones that actually are safe so the damage goes beyond the individual units they sell. There have already been publications in mainstream magazines and newspapers warning consumers about UV lights (the consumer markets are rampant with products that do not work, might be dangerous, and sometimes both). It may be difficult to move away from that image. Moreover, there is no guarantee that the cost-drivers for excimer bulbs will change as production scales, in which case it will not be affordable for a number of businesses. This is why we built the Airsafe. In theory, it's not as effective as irradiating the entire room -- in practice, your customers may feel safer knowing the UV light is well contained. I will be carrying around clear polycarbonate safety glasses if ambient UV light goes mainstream since those will block >99% UV.

See our other articles or take a look at our research paper summaries.

Learn more about the landscape of technologies marketed as making air safer: 1) Far UVC; 2) Radical Oxides e.g. Hydrogen Peroxide; 3) Ozone

Read the Reflow Labs Airsafe White Paper

See the key FAQ pages provided by a team of scientists on COVID-19 transmission and preventive approaches:

Learn more about our board of reviewers.

Get In Touch

Email or leave a message here to learn more about our products (>25,000 sqf at this time only, please).

Are you a small business or individual? Join our mailing list today and we'll let you know when our webstor for direct sales are live.

© Reflow Labs, LLC. All rights reserved.