House plants and indoor air quality

Let me begin with an admission.  I like indoor plants.  Throughout my career I have usually had one large plant in my work office.  Plants are nice to look at, and visitors often comment on the “nice plant.”  But while I like indoor plants, and suspect that they have some benefit in terms of psychological comfort to many, they do not appreciably improve indoor air quality any more than an old pair of socks that I would hang on a wall.  The sad thing about this never-ending saga is that those who are fooled into thinking that plants are good indoor air cleaners often avoid approaches and technologies that actually do effectively remove pollutants from indoor air.

In 2019, I gave a public seminar focused on myths related to indoor air quality.  I spoke about a lot of topics, from ion generators to essential oils. The audience was engaged throughout the seminar, asking many good questions.  But engagement quickly turned to enragement when I told them that plants do not clean indoor air. This issue is a visceral one for many on the side of purported VOC-munching house plants.  I have to believe that this response stems from years of repeated claims about the virtues of house plants as “air purifiers”, never with good evidence that can be translated to actual indoor environments.  There are some who grab on to something they saw on social media and communicate it widely without actually understanding the science.  And unfortunately, some researchers appear to seek a phyto-holy grail.  Many are responsible for bad research or flawed interpretation of results.  In all cases, none of these researchers are building scientists who understand spatial or temporal scale-up to actual building environments.

I recently did a Science Citation Index on peer-reviewed journal papers related to indoor plants for removal of VOCs or formaldehyde, with a lot of variations to try to capture as many papers as possible over the past 15 years.  I found 236 related papers.  Numerous papers have been published on pollutant removal by various types of plants.  Pollutants are dominated by volatile organic compounds (VOCs) such as toluene and benzene.  Formaldehyde, a ubiquitous indoor pollutant, has also been studied by many researchers.  So as not to be too critical, I will not mention specific authors here, nor will I mention journal editors who agreed to publish work of little relevance to indoor air quality.  Rather, I will generalize common problems with a large majority of papers.  Those with access to Science Citation index or other bibliographic databases can explore on their own.

For starters, let’s assume an 800 ft2 (74 m2) apartment with a ceiling height of 2.8 m (total volume = 208 m3) with a reasonable air exchange rate of 0.5/hr.  This translates to a volumetric flow rate of air into the apartment of 104 m3/hr (assuming similar indoor and outdoor temperatures and pressures, this is also close to the volumetric flow rate out of the apartment).  To reduce the concentration of a specific VOC by 50% (a measurable reduction) would require an air cleaner with a clean air delivery rate (CADR) of 104 m3/hr.  There are standalone air cleaners on the market that employ activated carbon and that can achieve this CADR for many VOCs.

So how does this relate to plants?  Well, here is the root of the problem (pun intended).  Research that has been completed on plants as air cleaners almost always suffers from one to all of these problems:

  1. Experiments are done in relatively small chambers that lead to MUCH higher plant (and potting soil) surface to chamber volume ratios than had the plant been placed in a large room, nonetheless an 800 ft2 apartment or even larger home.
  2. Experiments are completed with very chamber low ventilation rates, and some in completely sealed chambers.  Zero ventilation may be relevant to the international space station, but not to environments where nearly all of humanity spends the majority of its time.  Remarkably, many papers do not even state the chamber air exchange rate (ventilation rate), likely not recognizing the importance of this factor for back-calculating parameters that are relevant to assessing utility of plants in actual buildings. 
  3. Decay of pollutants injected into chambers is measured over long time periods, sometimes several days (required to actually measure decay).  Air generally stays in actual indoor spaces for tens of minutes to a few hours (not days), depending on air exchange rates of those spaces.
  4. Pollutant levels are often highly exaggerated by 100 to more than 1,000 times (or more) of what is typical in buildings.  This can affect the bioactivity of microbes in potting soil.

Consider the following statement in a recent paper … “A sealed perspex chamber with lid and fan was designed to ensure minimum leakage, proper aeration and distribution of benzene inside the chamber. Five different ornamental indoor plants were placed inside the chamber sequentially and exposed to a concentration of 5 ppm benzene for 30 h each. The leakage of benzene was checked beforehand.”

This type of approach is typical and of little relevance, unless the data are used to back-calculate parameters that can be scaled-up to actual buildings.  In this case, while spectacular results are presented, the scale-up of data to actual buildings shows these plants to be wholly irrelevant unless your home, workplace, or child’s classroom is packed with a deep jungle of plants, is extremely poorly ventilated, and trays of gasoline with an open surface are  emitting benzene into the indoor space.  I want to underscore the fact that this is not untypical of research done to show how houseplants can improve indoor air quality.  The research methods might have low experimental error, but the approach and outcomes are simply irrelevant (underscored x 100) to the real world.

What’s concerning about most research on houseplants is that the work is published without analysis of data relevant to actual buildings or insufficient data for building scientists to extract relevant parameters like clean air delivery rate (CADR) for the plant.  Extraction of such parameters is relatively easy when appropriate data are collected and analyzed using a basic concept in environmental and chemical engineering known as a mass balance (follow the mass).  To be clear, there are a few papers (a very small minority) with sufficient data and methodological description to allow determination of CADR.  For those papers, the back-calculated clean air delivery rate (CADR) for different types of plants and VOCs tends to be extremely low, e.g., 0.01 to 0.1 m3/hr (0.5 m3/hr would be very generous given published data).  To put this in another way, that 800 ft2 (74 m2) apartment described above would need to have somewhere between 1,000 to 10,000 plants in it to remove 50% of VOCs (perhaps 200 plants under the most generous of assumptions).  And yes, a larger home would require many more plants than this!

Here is how results in papers can be highly deceiving and what happens when one puts results in context.  Shown in the plots below are not data from an actual paper, but reflect approximate results from many studies.  Plot A shows the normalized concentration (pollutant concentration divided by its initial value after injection) plotted against time in hours.  In this case I have assumed a sealed chamber so that the entire decay can be attributed to pollutant removal by a plant in a 200 L chamber.  Approximately 95% of the pollutant has been removed over 72 hours.  That’s great, right?  Run out and grab a couple of potted plants for your home, office, or child’s classroom.  Not so fast.  The clean air delivery rate (CADR) back-calculated from this analysis is only 0.0125 m3/hr.  Compare this with, say, a CADR of 500 m3/hr (300 ft3/min) for portable HEPA filters that remove particles (4 to 5 orders of magnitude lower for the plant in this analysis).  Now, if one uses 0.0125 m3/hr in a mass balance to estimate the % reduction of a pollutant due to use of indoor plants in the apartment descried previously, we get plot B.  Seventeen plants buys you about a 0.25% reduction in pollutant concentration.  It takes over 80 plants in the apartment to get to a 1% reduction in pollutant concentrations (levels) and we do not even achieve 2.5% with nearly 200 plants.  Note that these concentration reductions are less than or approach the experimental error for specific VOC measurement and analysis, i.e., we would have no confidence that we could actually see a difference in pollutant levels, even with what would appear to be an apartment converted to a dense greenhouse.

If one wants to give plants the benefit of the doubt it is instructive to look at ozone, a pollutant that ought to be easily removed by chemical reaction when ozone molecules come in contact with plant surfaces or potting soil.  One research team (proudly from Portland State University) did very well-controlled experiments using five common indoor houseplants touted for improvement of indoor air quality.  They back-calculated appropriate parameters to allow a scale-up analysis for actual buildings.  Applying their results to a 1,800 ft2 home suggests over 90 houseplants are needed to remove somewhere between less than 1% to less than 10% of ozone.  Not to dwell too much on my old socks, but I could hang 90 old pairs around the house and see ozone removal benefits that might exceed those of the house plants that were tested based on reactions with squalene from my skin oil left on the socks.  In either case, there would be unwanted reaction products formed and released to indoor air. And that leads me to the topic of unintended consequences.

The jungle that we have created comes with some additional problems.  Plants emit sesquiterpenes and other easily oxidized terpenoids that can leave a long-lasting stench.  Those who have ever been inside a home that was once a marijuana grow house know what I am talking about.  It can be difficult to sell such homes.  And those terpenoids can be pretty reactive with a little ozone around, e.g., in the summer ozone season or if one also myths into using an ion generator as an air cleaning device.  Related reactions can lead to light and heavy carbonyls that can irritate the upper respiratory system or worse, and that can also linger on surfaces with a very slow release over weeks to months.  Furthermore, watering all of those plants will affect moisture balances in the indoor space, with possible mold problems as a result.

I have seen some propose the addition of granular activated carbon into the soil that supports the plant with forced air flow through the soil’s root zone.  This would allow VOCs to be adsorbed to the activated carbon and biodegraded by microorganisms in the soil.  This is really not as much of a houseplant as it is a more conventional packed-bed biofilter, which could work in theory.  However, even if a highly generous 100% of VOCs are captured in the soil bed, one would need to pass 100 m3/hr of air through the bed to get the same results as the commercial air cleaner described earlier.  That’s a lot of air flow. For an 8 inch diameter planter it equates to an air speed of almost 1 m/s through the bed.  Water would have to be replenished frequently due to evaporation, and things could get messy with moisture, soil particles, bacteria, and activated carbon released from the bed.  To avoid these complications a smaller volume could be used, e.g., 0.5 m3/hr, but that gets us back to a jungle of house plants. 

Any way you trim this (pun intended), using plants to reduce pollutant levels indoors is simply not practical.  Appreciate plants for their beauty, but don’t ask them to perform miracles.

Alas, it is time to go water my indoor plants.

The Dose Makes the (risk of) Infection

As the number of inhaled viruses or aerosol particles that contain the virus increases, the probability of infection increases. As such, we should be paying attention to inhaled dose, or better yet inhaled deposited dose, as a value that we want to reduce for scenarios involving two or more persons in an indoor space.

I have been speaking about the importance of inhaled deposited dose for several months. This very important concept seems like an innocent bystander to a lot of discipline-specific angst around semantics involving airborne particles, droplets, aerosols, close contact, etc. If a particle is in the air and you inhale it, it can lead to inhaled deposited dose. I do not care what anyone calls it or where the virus-carrying particle was inhaled. It may have been inhaled while challenged by a concentrated aerosol stream in close contact with an infector (near field) or at a lower concentration in the far field (see Figure 1). If it is inhaled it can deposit in the respiratory system. Period. Your immune system does not care what pathway it took to get there.

Importantly, concentration in air is NOT dose. Exposure is NOT inhaled dose. And more attention should be given to inhaled deposited dose (IDD) of aerosols, which after all are the effective rideshare for SARS-CoV-2 into your respiratory system. Why is IDD important? Because understanding the basic factors that influence IDD provides a roadmap for reducing dose and associated response (risk of infection). Further, once a dose-response relationship is developed we will need to be able to effectively model inhaled dose or inhaled deposited dose to use a dose-response relationship effectively. Such knowledge will help with strategies to prevent infection, as well as appropriate post-outbreak forensics. I have been spending my “free” time on development of emissions-to-IDD models.

What is the inhaled deposited dose? The IDD (Di) is the total number of particles in size range “i” that are inhaled and that deposit in a person’s respiratory system. In mathematical form, it is the product of four variables.

Di = Ci x B x t x fi

It is worth taking a closer look at each of these variables, as understanding the factors that affect these variables yields valuable insights into how to reduce Di, and therefore how to implement effective risk reduction strategies. 

The term Di corresponds to the total number of particles in size range i that actually deposit in the respiratory system, from the nose to the deepest recesses of the lungs, i.e., alveolar region. The units are simply number (#), as in number of deposited particles in a specific size range. A specific particle diameter is used to represent the range i and is often taken as the geometric mean of the endpoints of a small size range. For example, if a particle size range is 1 to 1.5 μm, the geometric mean is taken to be the square root of the product of these end points, or (1 x 1.5)0.5 = 1.22 μm. More on how we use this later. 

The term Ci is the concentration of particles in size range i in the breathing zone of a susceptible receptor.  This is taken to be a particle count in size range i per volume of air, e.g., # of particles per liter or air or #/L. This concentration depends on the number of particles emitted from an infector, which can vary significantly between infectors, by activities for the same infector, e.g., coughing or singing versus breathing while at rest, and as a function of time during infection. The value of Ci can be considerably lower if the infector is wearing a mask, especially in the near field if an infector faces the receptor, and even to some extent in the indoor far field. Further, Ci will be lower if the receptor is also wearing a mask (reduces concentration in breathing zone inside mask).

Greater ventilation reduces Ci, especially in the far field, by replacing indoor air that contains virus-carrying particles with outdoor air. It may also reduce Ci for close contact (near field) by enhanced mixing and dispersion of concentrated aerosol plumes emitted by an infector.

Values of Ci can also be effectively reduced by filtration (central or portable air cleaners). I have tweeted about nuances related to filtration and will not repeat that information here. Others have as well. None of this is new nor is it rocket science. 

Finally, we tend to focus on how much time we spend indoors to reduce our exposure. But remember that the amount of time that an infector spends in a space also affects Ci. If an infector walks into a pharmacy and spends 5 minutes there, the time-averaged concentration in the far field associated with their visit will be much lower than if they spend 30 minutes in the pharmacy. Small particles will accumulate in indoor air and, if the infector stays in the environment for a long time, may reach a steady-state concentration in which the rate of emissions equals rate of loss due to ventilation, surface deposition, filtration, etc. When the infector leaves the indoor space the concentration of virus-carrying particles will decrease at a rate defined by these same removal mechanisms. This may be very important for a scenario involving an infector who just left a classroom at the end of a lecture. Students entering the classroom immediately afterward will be exposed to a much higher value of Ci, and thus inhaled dose, than if they had entered an hour later. This is one good reason for leaving an unoccupied class period between each occupied period in a given classroom.

The variable B is the respiratory minute volume and has units of L/min of air inhaled. Its’ variability is often overlooked but can be very important. The average minute volume of an adult typically ranges from 5 to 10 L/min for rest to simple exertion, e.g., driving. But during heavy aerobic exercise, this value can increase to 50 to 60 L/min or more, up to 10 x or higher volumetric flow rate than what you are likely experiencing as you read this. This large range is why I have expressed such concern about gyms that include significant aerobic activity. Heavy breathing can significantly increase both emissions from an infector and inhaled deposited dose. We have heard a lot of concern about the risks of spending time in crowded and poorly-ventilated bars and restaurants, and for good reason. But crowded and poorly-ventilated gyms with those engaged in aerobic exercises might be even worse because of the elevated respiratory minute volume.

The variable “t” is a fairly obvious one and relates to the amount of time that a receptor spends in an indoor far field with an infector or an infector’s residual aerosol (after infector has departed), or in the direct near-field plume of an infector who is breathing, speaking, singing, coughing, etc., in your direction. For consistency with the units of other terms used here, time should have units of minutes (min). Minimizing time in indoor environments other than your own home is important. Do not go if it is not essential. If essential, go and figure out how to get out in the least amount of time.

The variable fi is the fraction of particles of a specific size range that deposit in the respiratory system (no units). Where particles deposit is likely very important. Particles can deposit anywhere from the nose (head region) to the deepest recesses of the lungs (alveolar region) and everywhere in between (largely the tracheobronchial region). Where a particle deposits depend on the mode of breathing (nose versus mouth), particle size, and to some extent the rate of breathing. Different models can account for these differences and can be used to predict particle deposition. Importantly, not all inhaled particles are deposited in the respiratory system and may come right back out during exhalation. For example, only about 45% of 1 μm (1 micron) particles will actually deposit in the respiratory system (mostly in the nose) for nose breathers doing light exercise (Figure 2). That means that 55% comes right back out in exhaled breath.

It is useful to convert the number of particles of various sizes deposited in different parts of the respiratory system into a volume, e.g., Figure 3 for a simulated 75-minute classroom lecture with an infected teacher. Results show the number of particles deposited in three major regions of the respiratory system of each nose-breathing student for different ventilation and control scenarios. The values at the top of each bar correspond to volume of particles deposited. Total particle volume can be approximated by assuming spherical particles and determining the volume of a particle in a size range using the geometric mean value described previously. This volume can be multiplied by the number of deposited particles in that size range to get the total volume of those particles. And the sum of volume over particle sizes yields total volume. If a viral load in number of viruses per mL is known, it may need to be adjusted for evaporation of water from particles when they are released to indoor air (an effective increase in viral load/mL of particles). This adjusted viral load can be corrected for the fraction of viruses that are infectious. 

As an example, assume that an IDD calculation leads to an estimate of 20 picoliters (pL), or 10-12 L, of particle volume deposited across all regions of the respiratory system. A viral load of 1011/mL is equal to 1014/L. If 1% of these viruses are actually infectious, the fraction infected is 0.01. This means that 1014/L x 0.01 = 1012/L are infectious viruses. This value can be adjusted based on virus inactivation rates and the time between particle release and inhalation. If we multiply 1012/L by 20 x 10-12 L for volume deposited we get 20 infectious viruses deposited. Of course, this assumes the same viral load across all particle sizes, and it is not clear that this is the case for SARS-CoV-2. However, it is not difficult to adjust the calculation if we someday have size-resolved viral loads and fraction infected.

I hope that this is a helpful blog. I have been developing models of different complexities that use this concept to illustrate how differences in building design and operation, human activities, and emissions affect inhaled deposited dose. The development and release of models have been a very slow process my day job is almost all-consuming. I am hoping to get one of the simpler models out soon with the help of some colleagues who are user-interface gurus.

Stay tuned and please venture back to this site from time to time. I am making a lot of my seminar, webinar, and lecture slides/notes available here, will continue to expand the list of links for other good information sites, will archive these blogs, and more. Please provide an acknowledgment at point of use if you do use any slides or other information from this site. Otherwise no charge.

Be careful out there.

We will get through this, but it will take a lot of patience. Keep your guard up!