Bedtime CO2 during wildfires & air pollution

Bedtime CO2 during wildfires & air pollution
Diagram from AC Infinity

Late summer in California increasingly implies preparing for wildfire smoke.

By “preparing” I mean focusing on indoor air quality, and in particular: Particulate Matter 2.5. We’re instructed to seal off our windows and doors from the outside and– if able– convert our bedrooms into clean rooms. The number of air cleaners (also known as air purifiers) on the market is on the rise.

Working to improve indoor air quality by mitigating PM2.5 is a good thing. But these efforts often ignore a key element of indoor air quality: carbon dioxide levels.

What’s the deal with CO2

Why does indoor carbon dioxide (CO2) matter? As we stay in an enclosed space, like a bedroom, the CO2 we exhale accumulates and we need to refresh the air to reduce its level. But when we seal up our rooms to prevent outdoor air from coming in, as is recommended in polluted / smokey conditions, we also reduce our access to necessary “fresh” air. Without a way to exhaust the accumulated CO2 and bring in new air, indoor CO2 levels rise.

After a particularly smokey evening with an AQI over 165, my partner and I woke up– sooner than intended– to a CO2 level of 3700 ppm [1]. We shrieked, arms and legs and sheets flying, as we ran over to let in fresh air. We had left our bedroom’s windows and doors closed, to ill effects.

Studies show that high levels of CO2 impact health and cognitive function. For example, in the 2016 CogFX study, Harvard’s Joseph Allen and collaborators showed that compared to a baseline of 550 ppm steady-state CO2, office workers’ cognitive function scores were 15% lower on days where they spent 5-6 hours in an indoor space at ~945 ppm, and 50% lower on days spent at ~1400 ppm. Subsequent studies have found similar cognitive impacts in subjects from metabolically-produced CO2, although not pure CO2 [2]. An Indoor Air paper on CO2’s impact to sleep and next-day performance also found logical thinking marginally better for metabolic CO2 levels under 850 ppm vs over 2000 ppm.

Ventilation, or the supply of fresh air and the exhaust of dirty air, has received more attention than usual these days due to the pandemic. Since people breathe out CO2, a high level of CO2 in a space indicates a greater possibility of airborne viruses present in the breathed-out air. The greater the ventilation, or air change rate, the greater the opportunity for fresh air to flush out contaminants. As we can see, we need increased ventilation and lower metabolic CO2 levels in our indoor spaces for both mental acuity and public health.

Unfortunately, mitigating the indoor accumulation of CO2 becomes difficult in wildfire season (and dealing with polluted outdoor air in general). We need to manage the tradeoff between reducing outdoor PM2.5 exposure WHILE ALSO reducing indoor CO2 exposure. How should we think about this tradeoff? And what are our options to manage it?

The tradeoffs of managing CO2 vs PM2.5 at home

I’ve known for a while the poor standards of CO2 in commercial buildings, but it wasn’t until the pandemic that my attention turned toward the CO2 in my own home. In early 2021 I bought a CO2 monitor to take around with me to public spaces. I was trying to build up an intuition for levels of ventilation (still a work in progress!), to better gauge risk factors from coronavirus transmission in different types of spaces. Up until late summer, I hadn’t paid much attention to the sensor near my bed… until the 3700 ppm event. That kicked me off on a journey: How should I manage CO2 in my home, especially when the polluted air outside renounces its role as a fresh source?

Dr. Elliott Gall, Assistant Professor of indoor air quality and buildings at Portland State University, shares the tradeoffs between managing indoor CO2 levels (e.g. bioeffluent exposure) while also trying to manage PM2.5 from wildfire smoke.

In general, there exists a trade-off between bioeffluent exposure and PM2.5 exposure in our clean room.

What about sleeping? You might consider leaving the door open at 50% for an entire 8-hour period. This will allow exchange of air between the bedroom and the rest of the home and strike a reasonable middle ground between PM2.5 and CO₂ exposure.

Gall shows that opening your bedroom door 50% during an air pollution event results in CO2 levels of ~1650 ppm and a PM2.5 exposure average of 21 µg/m³ (assuming two people in a 150 sqft room, 100 CFM air cleaner, and rest-of-house PM2.5 level of 117 µg/m³). “Not ideal,” Gall writes, “but during a wildfire, likely an acceptable outcome.”

Opening the door to 50% seems like a reasonable solution given the options, but personally (and assuming the ability to still keep PM2.5 levels low) I’m not satisfied with sleeping at a bioeffluent / CO2 level of over 800 ppm. As we saw from the studies earlier, cognitive function decreases by 945 ppm metabolic CO2. And in general, I don’t feel well-rested once those figures start to creep up [3]! (I lack evidence here though, so maybe I should study my heart rate variability and run my own cognitive tests. Can I integrate my CO2 sensor with my Whoop yet?) As air pollution events occur more frequently (or if I lived in a city where poor air pollution was a constant, like when I lived in Beijing in the early 2010s), this tradeoff becomes more persistent. There has to be a better way than “good enough.”

For all these reasons, my partner and I decided that we needed a way to actively reduce the CO2 build up in our bedroom, right away. But what were our options? Air cleaners and MERV filters help mitigate PM2.5 and VOCs, but not CO2. To address indoor metabolic CO2, short of a personal direct carbon capture system– like the lithium hydroxide canisters used on submarines and spacecraft– we needed a new, filtered supply of outdoor air.

Building a system to reduce CO2

We began to think through our options for a forced air system. (Unfortunately, indoor plants these days don’t cut it.) Our bedroom doesn’t have central ventilation, so we needed an isolated solution. We also already had an air cleaner (a la this homemade Corsi/Rosenthal Box) to help with the PM2.5. Given that we live in a temperate zone– hello SF Bay Area!–we didn’t care about managing sensible & latent heat (read: temperature & moisture) with a packaged unit like an air conditioner or heater, and we were also less concerned about heat recovery. (But if we were, and we had more $$$ to spend, we might have considered an HRV or ERV). Last but not least, we wanted to avoid chemical reactions… no accidental ozone creation please.

We gradually narrowed in on our desired solution: a small, modular, temporary system that could bring in outdoor air, filter it through a MERV 13 filter, and then blow it into our bedroom (ideally with minimal noise and energy use).

Of course, it makes sense where we found our optimal solution: the cannabis growing industry!

Figure 1. Ventilation 101 from AC Infinity (Source: AC Infinity). This image shows one of several possible configurations for a grow tent, care of AC Infinity. Shout out to their customer service team for answering my questions!
Figure 2. Our fresh air supply system. Our modular unit has an intake hose situated in the window, a filter box housing a MERV 13 filter, and an in-line fan that pulls fresh air from outside, filters it, then blows it out into our bedroom. (See below for more details on how we built this system.)

And now, the experiments

Once our unit was set up, it was time to run some amateur experiments. How well could our system manage our metabolic CO2? How did our unit compare to more passive activities, like opening doors or windows? And, which scenario best met our needs?

But first, some basics. Our bedroom is roughly 1600 cubic feet, about 200-sqft floor area with 8-ft ceilings. Our homemade system can supply air at multiple settings, and so we chose four levels to test: fan settings of 25%, 50%, 75%, 100%, equivalent to 30, 60, 90, 120 cubic feet per minute (CFM). (Even though the in-line fan is rated at 250 CFM, when it pulls through the ductwork and MERV 13 filter box it maxes out at 120 CFM.) We also tested two non-mechanical scenarios, a) our window opened 2.5 square feet, and b) our door opened 50%. Beyond the fan or prescribed window/door openings, the rest of the windows/doors were sealed. For each test, we (2 people) made sure to spend at least 7 hours overnight in the room. We also kept our CO2 sensor in the same location: near the bed.

Figure 3 shows all tested scenarios. Very quickly, we see the 3700 ppm event I mentioned earlier– with all windows and doors closed. The takeaway: actively providing fresh air, either through a supply fan or via more passive strategies like door or window openings has significant impact to accumulated CO2.

Figure 3. All scenarios. You can quickly see how small acts of supplying fresh air have major impact to accumulated CO2.

In fact, that no-windows-nor-doors-opened-at-all event (i.e. 0 CFM) overwhelms the graph. Let’s kick it out and focus on more active strategies.

Figure 4. Active CO2 management strategies. In our bedroom, the 60 cfm supply fan has a similar outcome to opening the window at least 2 square feet.

OK, that’s better. What do we see? First, we notice that turning on our fresh air supply at 30 cfm was equivalent to leaving the door open about 50%. (You’ll notice the door 50% curve builds up then drops over time; this is likely because it takes time for the air from our bedroom to mix with that from the hall; the door is farther away whereas the sensor is closer to us.)

Next, we see that running our fresh air supply at 60 cfm was equivalent to leaving a window open about 2.5 square feet on a moderate temperature day, though this surely would vary based on outdoor air conditions. In both cases, CO2 levels stayed below 800 ppm.

Finally, we see that running the fresh air supply at 75% to 100% (delivering 90 – 120 CFM) delivered the best results, at CO2 levels less than 600 ppm. (Note that this higher ventilation rate comes at a high noise cost.)

Where you personally land for your own sleeping health likely depends on your tolerance for CO2 and the level of effort you’ll take to get there. I prefer less than 800 ppm metabolic CO2 when sleeping, yet I don’t think the noise and unnecessary energy use from the fan on full blast is worth the further reduction.

What’s your preferred CO2 sleeping level?

From these simple experiments, I think I’ll stick with keeping the windows open in non-polluted times. But, on days when I cannot trust the outside air to be clean enough, I’ll plan to run our supply unit at 50% fan speed (60 cfm). (Of course while running our air cleaner, too.)

I hope that we pay more attention to the accumulated levels of CO2 in our bedrooms while sleeping, even in non-polluted times. There’s likely a reason your grandparents and their grandparents often chastised the youngsters for not “airing out” their homes enough. Over time our built environments have played strong (unfair) defense to the fresh air that revitalizes spaces within. It’s time to redirect our attention to its impact.

What are YOUR CO2 levels at night?

Building the Unit

Welcome to the how-to portion of this post. Here’s us trying to play Apollo 13. (Remember, the crew and their support teams were also trying to get the indoor carbon dioxide levels under control!)

Bill of Materials

Fresh Air Supply Unit (Total Cost = ~$212)



  1. Mount the window and duct adapter

2. Use the clamp to attach the hose (duct) to the adapter

3. Connect the hose to the filter box (plus make sure the MERV 13 is happy in there!)

4. And connect the in-line fan to the filter box as well

5. Voila. A modular system, ready when you need it


[1] 150+ AQI events sound terrible to me today, but keep in mind that Beijing’s annual average was 175 AQI from 2010-2014. (Refer to Beijing’s historical PM2.5 concentrations here ; you can convert them to AQI using AirNow.)

[2] It’s important to note that subsequent studies have found varying results compared to the CogFX study. For example, Rodeheffer et al did not find poor cognitive performance in submariners at even 15,000 ppm pure CO2 exposure. A 2019 paper in Nature by Ryder et al also did not find poor cognitive performance in astronaut-like subjects at pure CO2 levels of 2,500 ppm or 5,000 ppm (although they did find low performance at 1,200 ppm). Note that these subsequent studies often augmented metabolic CO2 with pure CO2 to achieve higher levels; Zhang et al found that cognitive performance decreased with metabolic CO2 and other bioeffluents, but did not decrease with pure CO2. Moreover, these subsequent studies utilized short exposure times, typically 2.5 hours, much less than what would be common in a sleeping environment. There’s certainly room for further investigation.

[3] I think this also means I wouldn’t do well in today’s submarines or the International Space Station, where indoor CO2 levels are reported to be around 5,000 ppm or higher! Shout out to my family members who’ve done this regularly over their careers.