the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 22 Aug 2025
The UBC ATMOX chamber: An 8 m3 LED-powered modular environmental chamber for indoor and outdoor atmospheric chemistry
Abstract. Environmental chambers are controlled reaction vessels used to investigate atmospheric processes such as photochemical smog, atmospheric fate of molecules and secondary organic aerosol formation. Environmental chambers are typically equipped with UV-A fluorescent lights with wavelengths between 350 and 410 nm or xenon lamps with wavelengths between 300 and 800 nm. However, these types of lights increase the temperature of the chamber, are energy intensive and are not tunable to specific wavelengths. Fluorescent lights are also becoming redundant in the light industry. To address these issues, we prototyped the use of light-emitting diode (LED) lights from Violumas on our 8 m3 environmental chamber for photochemical experiments to enable stratospheric, tropospheric and indoor light conditions. The University of British Columbia (UBC)'s Advanced Techniques for Mechanisms of OXidation (ATMOX) chamber was assembled with custom-made wide-angle LEDs of six different wavelengths from Violumas: 275, 310, 325, 340, 365, 385 nm. We also added LED grow plant lights (Feit Electric) for irradiance between 450 and 630 nm. The LEDs were wired to a potentiometer control panel to modulate their output on a per wavelength basis. We used a total of 1440 custom LEDs and 1320 commercial grow plant LEDs, costing USD$ 44,951 and USD$ 1,300, respectively. We demonstrate their energy efficiency, their ability to generate less heat, and their ability to generate wavelength-specific photochemical processes. Furthermore, chemical actinometry using NO2 enabled us to calculate a photolysis rate constant (JNOx) ranging from 2.28 × 10-4 to 4.93 × 10-3 s-1, which is nicely comparable to 4.50 × 10-3 s-1 in Vancouver, Canada during the summer solstice.
In addition to the lights, the UBC ATMOX chamber was designed to be particularly modular. The chamber frame has 12 aluminum T-slot rails (2.66 × 2.66 × 3 m, 80/20 Rocky Mountain Motion Control), and a pulley system to enable the 8 m3 bag to collapse and inflate, to perform batch or continuous mode experiments. The Teflon chamber bag has sealable openings at each corner to allow access to the interior of the bad for regular thorough cleaning. Overall, our chamber is allowing us to study topics of current interest in atmospheric chemistry: from the fate of indoor air fragrances to cannabis emissions and from wildfire aerosol photochemical changes to the biogeochemical cycling of selenium.
- Preprint
(4675 KB) - Metadata XML
-
Supplement
(10105 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-3041', Anonymous Referee #1, 18 Sep 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-3041/egusphere-2025-3041-RC1-supplement.pdfCitation: https://doi.org/
10.5194/egusphere-2025-3041-RC1 -
RC2: 'Comment on egusphere-2025-3041', Anonymous Referee #2, 20 Sep 2025
General comments
Lee et al. describe the setup of a new environmental simulation chamber with collapseable Teflon walls and flexible illumination by different types of modern LED. The chamber design is described in detail with emphasis on the LED choice, setup and operation. The overall concept is sound and the advantages of using flexible LED illumination are convincing. Also, the comparison of basic properties of this new chamber with several existing facilities worldwide is useful. However, the paper is not well structured. Sections 2 and 3 contain partly redundant information and recurrent topics. Many technical details could be shifted to the supplement to improve the relevance for the scientific audience.
My main criticism refers to the analysis of the obtained data which is insufficient. The authors determined jNO2 in actinometric experiments but the relationship between the measured LED irradiance spectra and the actinometric jNO2 is merely discussed on a qualitative level. What these results mean for the photolysis of other species that will sooner or later be used in this chamber remains unclear. The authors do not seem to be aware of the difference between spectral actinic flux and spectral irradiance or do not discuss it adequately. These quantities are not exchangeable as implied by the jNO2 determined from the outdoor irradiance measurements. Moreover, there is an inconsistency in the actinometry data where much more O3 is formed than NO in the photolysis of NO2. The paper therefore requires major revisions at several points as outlined in the specific comments below.
Specific comments
Line 11-12, lines 169-174: The focus on the costs of the LEDs is unusual for a scientific paper. As the authors note, the LED industry is rapidly evolving, and the price information will soon be outdated. Moreover, compared to the total costs of the chamber (including construction work and other instrumentation for useful experiments) the share of the LEDs is probably not decisive. In my view, the information given in Tab. 2 is sufficient.
Line 14: The JNO2 value of 4.5×10-3s-1 likely refers to clear-sky, local noon conditions which should be specified. However, for such conditions the value is too small by a factor of about 1.8 (dependent on aerosol load). Apparently, it is based on outdoor spectral irradiance measurements as described in lines 333-337 which explains the difference. I suggest that the authors access relevant literature to revise their approach (e.g. Hofzumahaus et al., 1999, https://doi.org/10.1364/AO.38.004443). Solar spectral irradiance and spectral actinic flux are highly correlated but not identical. There are empirical relations that can be used to convert these quantities (McKenzie et al., 2002, https://doi.org/10.1029/2001JD000601) but for the purpose of this work (estimation of typical outdoor values) a radiative transfer calculation would be sufficient. Also note throughout the text that jNO2 is formally the (first-order) NO2 photolysis rate coefficient while the NO2 photolysis rate, like other reaction rates, is the product of jNO2 and the NO2 concentration.
Sections 2.1.1 and 2.2.1-2.2.4 contain too many technical details for the scientific reader. The information may be useful for someone who wants to set up something similar, but it’s sufficient to make these details available in the Supplement.
Sections 2.1.2, 2.1.3, 2.2.5, 2.2.6 could be merged with Sect. 3.1, 3.2 and 3.4 to obtain a more concise description of the chamber concept, instrumentation and properties. An information that is missing is whether the content of the chamber was stirred during experiments. The volume is not illuminated homogeneously and dependent on the lifetime of the species of interest, concentrations gradients may build up that can influence measurements and their interpretation. Both approaches, batch and continuous flow, require well-mixed conditions. This ideally also ensures that measured concentrations will not depend on the position of inlet lines and that flushing of the chamber results in a predictable dilution.
Section 2.1.3. The description of the cleaning procedure is tenuous. The authors should explain how the air used for flushing was cleaned or if commercial synthetic air with a specified purity from cylinders was used. Moreover, “particle-free” conditions are unrealistic and the minimal NOx concentration that can be reached should be specified. The cleaning topic is addressed again in Sect. 3.4.3 but remains inconclusive. What exactly means flushing over night? How often was the content of the chamber exchanged? Did it include collapsing the chamber to speed up the exchange? And what means rinsing the bag with water? Is that done manually or automatically?
Line 212: “The spectrometer was calibrated…” Was this a factory calibration or a calibration performed by the authors? Provide more details also on the spectral resolution.
Line 213-215: Note that by the described measurement configuration the irradiance from the plant grow LEDs is not adequately captured. In the extreme case of an LED sitting in the middle of the chamber ceiling, nothing is measured because the cosine receiver is blind for an incident angle of 90°.
Eq. 1 - Eq. 3: I couldn’t find Eq. 1 in Moreno and Viveros-Méndez, 2021. And I couldn’t find the reference Dragomir et al., 2014 at all. Give more details in S2. Perhaps also define the Rk used in Eq. 2 in a separate equation and insert Rk as the denominator in Eq. 3. However, the question is, how useful the prediction of irradiances is to characterize the conditions during photochemical experiments. If you skip the cos(θ) factors you would at least get an idea about how actinic flux densities are distributed. This approach could also cover the plant grow LEDs adequately in your model which were not considered at all later in Fig. 8 and Fig. 9. Still this model will lead to differences with the actual chamber situation because of (possibly wavelength-dependent) scattering processes at the chamber walls that are not included. Moreover, the curtains seem to be reflective which will lead to an internal enhancement of radiation.
Line 234: The review by Rabani et al. is concerned with liquid phase actinometry and does not describe the determination of jNO2 as implied. The jNO2 approach was applied in the SAPHIR chamber before and the relevant publication is cited in the Supplement S3.
Line 242: The recommendation by Atkinson et al., 2004 is outdated. Current recommendations by NASA-JPL and IUPAC result in about 10% greater rate coefficients for this reaction around room temperature. Use proper units and indicate them. −10.89 is in kJ mol-1 and using R in units kJ mol-1 K-1 is uncommon.
Fig. 4: The 450 nm irradiance is probably too small for geometric reasons as explained above. Moreover, the 450 nm spectrum in the lower panel does not fit to that in the upper panel. In the caption, what do you mean by “…the difficulty in taking a completely dark blank outdoors”? I assume you refer to a stray light issue during daytime. A dark spectrum can always be measured.
Fig. 5, Fig. S9 and Tab. S2: How can you produce more than twice as much O3 than NO in the photolysis of NO2? There must be something wrong here. In the last three lines of Tab. S2 NO and NO2 seem to have been confused. That considered, it leads to a similar mismatch between NO and O3 also for the fluorescent light experiments. The jNO2 derived from these data may be incorrect.
Line 318: The citation Hawe et al., 2007 is improper. These authors didn’t investigate the UV-C absorption of NO2.
Lines 321-323: The meaning of this sentence is unclear.
Lines 324-326: The statement is misleading. As shown in Fig. 6 the absorption cross section of NO2 in the wavelength range of the plant grow LEDs is still about half of the values around 385 nm. The reason why NO2 does not photolyze above 420 nm is a drop in the quantum yields above 400 nm. The term quantum yield does not appear in the whole paper which is surprising because it’s key to calculate jNO2 from the measured spectra.
Lines 333-337: See my comments on the outdoor measurements above (abstract, line 14). To clarify: In contrast to atmospheric measurements the use of irradiance sensors (cosine receivers) is feasible to characterize artificial light sources like LEDs in the chamber. If the spectrum of the radiation that is emitted by a specific type of LED is not altered by reflections on the chamber walls, it can be assumed (and confirmed by measurements) that the spectra measured at different positions and viewing directions in the chamber are all the same on a relative level. The same spectrum then also applies to the shape of the mean actinic flux spectrum. All you must do is to scale the relative spectrum to an absolute level that results in the measured jNO2 from the actinometry. This “calibrated” spectrum is then applicable to calculate the j-values of other species as well, based on their absorption cross sections and quantum yields.
However, this procedure does not work if you combine different types of LEDs unless they are equally distributed and have the same geometrical emission (and reflection) characteristics. This is not the case here as shown in Figs. 8 and S9. You therefore must characterize the output of each LED type separately by actinometric experiments. The resulting calibrated spectra can then be added to a total spectrum dependent on the desired experimental conditions as indicated in the upper panel of Fig. 4. jNO2 actinometry seems to work for the 310-385 nm LEDs but not for 275 nm and 450 nm for which other species must be found. 2-nitrobenzaldehyde is not a good choice (see below).
Another relevant question is if jNO2 or other j-values are expected to remain constant while the chamber is collapsed in the batch operation mode.
Sect. 3 and Tab. 3 are confusing. The power consumption of other chambers are not listed in Tab. 3. The total power consumption of all LEDs is 10-fold compared to the fluorescent lights. Also the jNO2/W metric is in favor of the fluorescence lights which is not further discussed. Why is jNO2 for the indoor configuration with less LEDs and less power/bulb almost the same (and much more effective) that the tropospheric? And what irradiances are listed? Those measured or modeled in the middle of the chamber? The uncertainty estimates of 1% or less for the jNO2 are unrealistic.
Line 362ff: The reasoning for using 2-nitrobenzaldehyde is misleading. This compound is of no interest in the stratosphere as implied by the text because its tropospheric lifetime is very short (also Kahnt et al. do not discuss any stratospheric relevance of 2-nitrobenzaldehyde). On the other hand, stratospheric relevance is no precondition: you could use any compound that is photolyzed at 275 nm with a suitable j-value. The problem with 2-nitrobenzaldehyde is that the quantum yield of photolysis is poorly known. The measured photolysis rate coefficient does therefore not sufficiently characterize the chamber under illumination at 275 nm. For example, if you want to do an experiment under “stratospheric” conditions, the presence of ozone may be required but the photolysis rate coefficient of ozone with the 275 nm LEDs can only be roughly estimated based on the results obtained with 2-nitrobenzaldehyde.
Figure 7 needs more information in the caption: what instruments were used? How much of the VOC was injected? There is no shaded area indicating the illumination period. I assume it was between about 9:35 and 10:35 when the decay was faster. When this assumption is correct, the following questions arise: (1) What explains the rather quick decay before illumination? (2) Why doesn’t the experiment start with “particle free” air? (Sect. 2.1.3). The particles seem to increase in size during illumination while new particles were formed after the lights were off. The y-axis unit in the lower panel should be a concentration like ppb or a count rate proportional to the concentration.
Lines 370-374: This paragraph should be phrased with more caution. LEDs emitting radiation at 275 nm may be helpful to simulate stratospheric conditions in chamber experiments. But they do not reproduce the full solar UV-C range. Moreover, stratospheric conditions are characterized by pressures well below 200 mbar and temperatures in a range 200-260 K which cannot be provided here.
Section 3.4.1 is unclear. What do you mean by "temperature increase per joule" and how was it calculated? Tab. 4 tells me that with your indoor, tropospheric and stratospheric conditions you get greater ΔT/t compared to the fluorescent setup. The energy E spent for the fluorescent experiment is 512 W x 3360 s = 1.7×106 J. How do I get from this to a ΔT / t / E of 22.9 ? Apart from that, makes this comparison sense at all because the setups are completely different.
Technical comments/typos:
Line 19: Replace “bad” by “bag”
Line 79: The volume range 280-370 m3 is unclear. In Tab. 1 this chamber is listed with 270 m3 which corresponds to the value given by Rohrer et al., 2005.
Line 218 and 230: Probably Fig. S9 instead of Fig. 11.
Line 207: Fig S3 is the wrong reference.
The title of the paper in the Supplement is different from the main text.
Citation: https://doi.org/10.5194/egusphere-2025-3041-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,799 | 44 | 12 | 1,855 | 16 | 31 | 27 |
- HTML: 1,799
- PDF: 44
- XML: 12
- Total: 1,855
- Supplement: 16
- BibTeX: 31
- EndNote: 27
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1