3D Printing, Ultra Fine Particulates + Volatile Organic Compounds, and Should I Be Afraid?+Jan Wildeboer
posted about the recent news raising concern in the 3D Printing community regarding UFP and VOC emissions from various types of fused-deposition modeling based 3d printing - i.e. those forms of 3d printing that involve melting plastic, squeezing it through a hole, and laying it down in order to build up a 3d object.
I posted a dismissive response about the concerns, having recently heard a claim that emissions from candles were higher. My dismissiveness was on the basis that if candle emissions were substantially higher, and there was not a noticeable gap in longevity between practicing Catholics and other actively religious persons whose devotions involve frequent use of candles, and those who don't
come into the vicinity of candles regularly, then obviously the effects must be below the threshold where they would make a material impact on my life.
I was challenged on this by another commenter. When you are challenged in something you've posted, the temptation is certainly there to simply double-down, dismiss criticism, throw out an authority, or the like. Ideally, though, we should take a moment, at least, to consider, is the claim that I made reasonable,
and do I have good reason to believe it to be true?Reasonableness
This is the level of back-of-the-envelope math, or mental reasoning with bare orders of magnitude. If the claim is obviously wrong on its face, you fail reasonableness. If it is not obviously wrong, but requires extra assumptions, you might be in trouble here, and that should push you to find even stronger evidence. In this case, testing reasonableness went along the lines of - what am I comparing, and is it reasonable that they should even be comparable?
In this case, in one corner, we are applying heat and melting plastic. On the other, we are applying open flame, and melting wax, and burning a wick. So on the face of it, they seem at least comparable, and while it could swing either way depending on composition, it is at least not absurd that wax+flame+wick could produce more than plastic alone. Both, in these contexts, involve small amounts of mass being processed per unit time, so scale wise there doesn't seem any great problem. Did I have a good reason
I failed here. What evidence did I have for the basic claim, that candle flames emit more particles than FDM? I had heard someone say so. Given my environment, that someone was almost certainly a podcaster, and therefore not an expert. They undoubtedly were quoting someone else. But I could not put a name to the podcaster, much less to their source, so I had no basis for my claim at all, beyond hearsay. So I'm wrong then?
Well, no. I did not have a good reason, but that does not mean such a reason does not exist. I looked on this instead as an opportunity, a chance to confirm or deny my belief. After all, as someone with a 3d printer, I certainly have a dog in this race, and as a scientifically minded person, I would rather know the truth than successfully defend a falsehood.
So, I started looking. The original paper was of course linked by the original G+ post, by Jan, however, that link did not provide a numerical value. So I ran a search on Google Scholar, and found a 2015 paper whose abstract contained the numbers I desired. It gave me a figure for ABS particulates. Then I went looking for a paper on candles... I found that too, after a fair bit of digging to find the full text, the abstract being lacking in numbers. The units don't match!
One paper gave a value in ea/min (each/minute, or particles per minute), the other gave the emissions in micrograms per gram burned. How can I compare apples to apples here? I turned to WolframAlpha, an excellent online computational engine that supports units, and massaged the data I had, working it towards the form I needed. I used a graph in the paper to judge the size of the most frequently occurring particle, which appeared to be about 50 nanometers, and the lists to determine the form of the same particle, n-hexacosane, then used the open-source chemical modeler Avogadro, with assistance from PubChem via Google to obtain the mass and SMILES formula for hexacosane (CCCCCCCCCCCCCCCCCCCCCCCCCC), to compute an approximate length of the single molecule. Dividing the size by this value gave me a candidate molecule, 15-hexacosane, to use as a proxy for the bevy of particles emitted by the candle. Then I could take the emissions rate in grams and use the mass from PubChem and Avogadro's constant to compute a number of particles for that mass, to get a final answer in terms of particles per minute.I goofed, three times, Frink saved me
My first run using just WolframAlpha led me through all the calculations to get an answer, and I put all the numbers in my post so anyone could follow the trail. But WolframAlpha is ephemeral. When I went back later to write this article, I started by redoing the calculations in Frink, a freely-available units-aware programming language. (Mathematica, MapleV, and other such symbolic math software would have served as well.) I did this to record each of the values and computations so that they could run freely, the output of one fed into the next, with no room for a change of units or a mis-transcribed digit to hide. And immediately I noticed that the very first calculation was giving a different answer - I missed at first that this was due to a mistranscription and the original answer
had been correct, so that was both the first and second goofup. And third, I noted that I had missed a step in going from grams to particles, as I was going from grams burned in one minute instead of grams of particulates emitted in one minute.Final result
// Source: Emissions of Nanoparticles and Gaseous Material from 3D Printer Operation, Yuna Kim, et. al., Environ. Sci. Technol., 2015, pp 12044-12053
ABS_particulates := 1.61e10 1/minute
println["ABS particulate emission rate : " + (ABS_particulates->"minute^ - 1")]
// Source: Characterization of Fine Particle Emissions from Burning Church Candles, Philip M. Fine and Glen R. Cass, Environ. Sci. Technol., 1999, pp 2352-2362
// Data from CAN-4D:
burn_time := 15 min
amt_burned := 1.12 milligram
burn_rate := amt_burned / burn_time
org_compound_emission_rate := 1040 µg / g
println["Candle burn time: " + burn_time]
println["Candle mass burned: " + (amt_burned->"µg")]
println["Candle burn rate per minute: " + (burn_rate->"µg/min") ]
println["Organic compound emission rate: " + (org_compound_emission_rate->"µg / g")]
organics_emitted := burn_rate org_compound_emission_rate
println["Organic emissions per minute: " + (organics_emitted->"µg/min")]
// This value is estimated from the graph of particle size versus quantity produced
// #### #
// #### ##
// ### #
// ## ##
// 0.01 0.1
// It is on a logarithmic scale, so the peak at about two thirds of the way between 0.01
// and 0.1 is about 0.05
particle_size_mode := 0.05 µm
// Most frequently occurring particle was n-hexacosane
// Source: Avogadro (chemical modeling program)
hexacosane_length := 32.975 Å
num_chain_elements := floor[(particle_size_mode / hexacosane_length)]
println["Model molecule is " + num_chain_elements + "-hexacosane"]
hexacosane_mass := 366.422552 g / 6.02e23
candle_emission_rate := organics_emitted / (hexacosane_mass * num_chain_elements)
println["Candle emission rate is " + (candle_emission_rate->"min^-1") ]
println["Candle emission rate is " + (candle_emission_rate / ABS_particulates) + " times the ABS emission rate."]_Output
ABS particulate emission rate : 1.6100e+10 minute^-1
Candle burn time: 900 s (time)
Candle mass burned: 1120.0 µg
Candle burn rate per minute: 74.664 µg/min
Organic compound emission rate: 1040 µg / g
Organic emissions per minute: 0.077652 µg/min
Model molecule is 15-hexacosane
Candle emission rate is 8.5050e+12 min^-1
Candle emission rate is 528.27 times the ABS emission rate.Conclusion
My first value, which was of order 10^15, is probably not far off the mark for total particulates, though some of these would large enough to fall in neither camp. The adjusted figure of order 10^12 for just VOCs is still higher than total combined particulates for ABS by two orders of magnitude. So my podcaster informant probably had a good source, and now I have a reasonable claim and good reason to believe it to be true, which lends support to the assumptions underlying my hypothesis that significant risk from these sources would have previously manifested itself, and that I can therefore feel comfortable in taking reasonable precautions and not abandoning my hobby entirely.
And maybe I'm just weird, but yes, I actually had fun in the hour or so I spent yesterday evening researching and writing my response, in following up and finding the relevant research, in figuring out the path from the values reported in one publication to obtain values directly comparable to those of the other, and even in writing out the code form and finding my errors the next day. None of it was hard, I didn't have to actually do any
of that math on paper or in my head, and while there may be no good reason for anyone to trust my research or math, I left all the pieces there for anyone to check it for themselves, so my lack of authority in the matter should have no bearing.But what about Cyanide?!
The LC50 of hydrogen cyanide, from www.atsdr.cdc.gov
(Agency for Toxic Substances and Disease Registry), is 524 ppm for a 10-minute inhalation exposure, and death has been reported after as little as 135 ppm. Levels of between 5 and 15 ppm have produced symptoms in persons with occupational exposure (i.e. repeated exposure over years). What is a part per million? A milligram per liter.
ABS melt density - 0.97 gr/cc according to a quick Google search, times 4 cubic millimeters per second as an average printing speed, 3.88 mg/s. So, if the entire output was being sublimated and going straight to hydrogen cyanide (which is obviously waaay
overstating the case) it could be effectively adding 3 parts per million to a cubic liter of air around the printer. A small room, 8 x 10 x 10 ft, again going with a conservative case, gives 22,653 L. So, assuming any degree of reasonable airflow and mixing, and no
ventilation at all, you've got 22,653 seconds before each liter has that 3 ppm. That's six hours. Twelve hours to get to 6, where we start to see long term effects - and that was assuming total conversion to hydrogen cyanide! Once again, back of the envelope calculations to the rescue - as long as I don't stick my nose in close and sniff deeply, I am very unlikely to face significant cyanide poisoning from my printer, even without considering the question of whether and how much is released at normal temperatures, as opposed to from burning ABS.