Dataset: Primary results from the design and validation of an LED-based photochemical reactor

All methods are described in Ward et al., 2021. A summary is provided below. This study was conducted in the Ward Lab at Woods Hole Oceanographic Institution in Woods Hole, MA, USA between 2018 and 2021.

Description of the LED Reactor Assembly:
Each LED reactor is comprised of an inner housing, containing the samples, and an outer housing, on which is affixed an LED chip. On top of the outer housing is a washer that provides the base for the LED chips, which are mounted to a printed circuit board and a heat sink. Passive cooling by the heat sink is sufficient to maintain the samples at room temperature, 22 ± 1 °Celsius, without any additional cooling. The LED chips face down toward the inner housing which contains four quartz tubes. The inner housing is machined specifically for the dimensions of the tubes but can accommodate a variety of reaction vessel shapes and sizes depending on the reaction of interest. Both housings are painted black matte to minimize light scattering and to decrease stray light in the laboratory. A separate power supply is used for each reactor.

Spectroradiometry and Chemical Actinometry:
Spectral irradiance measurements in all reactors were made with a NIST-calibrated Black Comet spectroradiometer equipped with a cosine corrector (StellarNet Inc.). Chemical actinometry was assessed in the 309, 348, 369, and 406 nanometer (nm) reactors by measuring the hydroxylation of benzoic acid to salicylic acid in the presence of nitrite. Actinometry was not assessed in the 275 nm reactor because the apparent quantum yield (AQY) of salicylic acid production has not been reported at that wavelength. Salicylic acid production was quantified fluorometrically using an Aqualog (Horiba Scientific).

Photochemical Oxidation Experiments:
Photochemical oxidation (dissolved O₂ consumption) experiments were conducted as an example test of the LED reactors' long-term precision in measuring AQY spectra and oxidation rates. These experiments used Suwannee River natural organic matter (SROM; 2R101N) acquired from the International Humic Substances Society (http://humic-substances.org/). In total, six experiments were conducted over a six-month period using four replicate solutions of SROM. Triplicate experiments were conducted for one SROM solution on three consecutive days (referred to as experiments 2a, 2b, and 2c). The coefficient of variation (CV) of photochemical oxidation rates was 11%.

Before each experiment, solutions of SROM were prepared in Milli-Q water (Millipore) at a concentration of 20 milligrams (mg) SROM per liter (L) (~10 mg C L⁻¹). The solutions were adjusted to pH 7.0 ± 0.1 and allowed to equilibrate on a stir-plate for 24 hours prior to filtration with a 0.2 micrometer (µm) Sterivex filter (Millipore). SROM was then transferred to Milli-Q rinsed quartz tubes (15 millimeter (mm) outer diameter, 100 mm length; Technical Glass Products, Inc.) and sealed with a Viton-lined GL-18 cap with no headspace. The tubes were placed vertically in the inner housing with the flat bottom facing the LED. To minimize the impact of photon dose-dependence (i.e., the change in AQY as the samples absorb more light over time), we adjusted the emission intensity of the LEDs via the power supplies such that the moles of photons absorbed by SROM under all LEDs were equal within 5% (Figure 2b and Table S2 of Ward et al. (2021)). The total number of moles of photons absorbed by SROM was calculated by multiplying rates of light absorption (Qₐ; mol photons m⁻² s⁻¹, determined via equation 1 of Ward et al. (2021)) by the exposed surface area of the quartz tubes (1.1 cm²) and the time of the light exposure (43,200 seconds).

The incident photon spectral irradiance reaching the quartz tube (mol photon m⁻² s⁻¹ nm⁻¹) is determined with 1-nm resolution and the summation is performed with 1-nm increments across the relevant wavelengths for each LED (e.g., 332 to 376 nm for the 348 nm LED). Naperian absorption coefficients of SROM (m⁻¹) were measured using a Perkin Elmer 650 spectrophotometer and calculated as the geometric mean of the light-exposed and dark control treatments to account for photo-bleaching, which was minimal in all experiments (< 5%). The pathlength (z) was 10 cm, equivalent to the height of each vial. Photochemical O₂ consumption was calculated as the concentration of dissolved O₂ in the dark-control tubes minus that in the light-exposed tubes using membrane inlet mass spectrometry (Bay Instruments, Inc.).

Apparent Quantum Yields:
Apparent quantum yields (AQYs) for the photochemical oxidation of SROM (mol O2 mol photons⁻¹) were calculated for each tube in each LED reactor by dividing the moles of O₂ consumed by SROM by the moles of photons absorbed by SROM. The AQY data were fit to an exponential curve with 1-nm increments.