Up in Smoke: Most Aerosolized Fe From Biomass Burning Does Not Derive From Foliage

Iron (Fe) is a limiting micronutrient in many marine ecosystems. The lack of sufficient Fe can stunt marine productivity and limit carbon sequestration from the atmosphere to the ocean. Recent studies suggest that biomass burning represents an important Fe source to the marine environment because pyrogenic particles have enhanced solubility after atmospheric processing. We examined foliage representative of four distinct biomes subject to frequent burning events, including boreal/temporal forests, humid tropical, arid tropical, and grassland. We burned these samples in the absence of soil to isolate the Fe from the fine particle (PM2.5) fraction that is derived directly from the burning foliage. We find that <1.5% of the Fe in plant matter is aerosolized throughout the burn in the fine fraction. We estimate that between 2% and 9% of the Fe released from biomass burning can be attributed to the fine fraction of the foliage itself, and <50% from the foliage overall. Most of the Fe aerosolized during biomass burning is accounted for by soil‐suspended particles.

from plant material without inputs from uplifted soil.In brief, their approach calculated the Fe emitted from biomass burning by estimating the fraction of fuel consumption based on a global 0.1° × 0.1° fuel data set, the completeness of plant consumption, the average Fe concentration in each given biome, and the amount of Fe lost to the residue ash phase.The authors of this study found that the fine fraction (PM1 in their study) of the plant matter accounted for only 17 Gg yr −1 while the coarse fraction accounted for 460 Gg yr −1 of global Fe (Wang et al., 2015).To our knowledge, this estimate has not been corroborated with field data.
Here, we provide direct measurements of the fine fraction of Fe (PM 2.5 ) derived from burning plant material that is isolated from uplifted soil particles and compare our results to those ascertained using a mass-balance approach (Wang et al., 2015).We conducted small-scale biomass burning experiments of isolated plant matter because field studies from large burns cannot disentangle burning plant matter from soil-suspended particles.These experiments allow us to determine the amount of fine particulate Fe released directly from plant material.We then use a ratio of coarse-to-fine particles to estimate the amount of coarse particle Fe released from foliage (PM > 2.5; Ito, 2013;Luo et al., 2008;Wang et al., 2015).Our experiments suggest that the influx of plant-derived Fe is small relative to soil-suspended particles, but is not necessarily insignificant to the marine Fe budget.One study suggested that the fine fraction of pyrogenic particles has a solubility of 33% and the coarse particles have an initial fractional solubility of 4% (Hamilton et al., 2021;Luo et al., 2008).Thus, these fine combustion particles will likely have higher solubilities when deposited in the surface ocean.Thus, as a first step, it is important to quantify the amount of Fe (PM 2.5 ) mobilized from plant material during biomass-burning events.
We provide a first-order estimate of global plant-derived Fe from biomass burning by burning plants representative of distinct biomes.A significant limitation of previous work is that each study could only focus on a single combustion event (Kurisu & Takahashi, 2019;Paris et al., 2010;Siefert et al., 1996) and therefore could not provide a global perspective or differentiate among different types of biomass burnings (e.g., reed, savannah, pine, etc).Our study investigates vegetation representative of four regions: temperate/boreal, humid tropical, arid tropical, and grassland.We use the results from these experiments alongside estimates of the areal extent of biomass burning in each region to estimate a global flux of Fe from combusted foliage during biomass burning events.This study uses field data to estimate how much Fe comes from plant material versus lofted soil particles.Disentangling these two sources has important implications for estimating how much of the Fe deposited in the marine ecosystems may have enhanced solubility.

Materials and Methods
The foliage that was collected represents various biomes, including temperate, humid tropical, arid tropical, and grassland.The foliage was burned, and the particulate matter was collected on trace-metal clean cellulose filters mounted on a modified high-volume sampler (Figures 1-3).Briefly, the sampling stage was specially designed with polytetrafluoroethylene (PTFE, Teflon) and placed on top of the steel mesh of the sampler to prevent contact between the filter and the impactor (Figure 3).

Reagent and Material Preparation
Laboratory procedures were conducted in a Class 10 laminar flow hood at Arizona State University.Unless otherwise specified, acids used in this study were purified using a sub-boiling in-house distillation, and concentrated acids were diluted as needed with >18.2 MΩ-cm de-ionized water (Milli-Q).Plastic materials were degreased in a Micro90-trace metal cleaning solution (1%), rinsed with de-ionized water, soaked in reagent grade nitric acid at room temperature (7.8M), HCl (6.1M), and rinsed with Milli-Q water (Howard & Statham, 1997).Polytetrafluoroethylene (PTFE, Teflon) containers underwent an additional cleaning step by refluxing with distilled concentrated HCl on a hotplate for 2 hr.
Whole and slotted cellulose filters (203 × 254 mm (TE-241), 152 × 152 mm (TE-230-WH), Whatman 41, Tisch Env.) were soaked overnight in reagent grade HCl (0.24M) and subsequently rinsed with Milli-Q.This entire process was repeated for a total of three wash cycles.Filters were allowed to dry before being stored in plastic zipper bags.

Sample Collection
Foliage samples were selected to represent a global range of plants subject to biomass burning (Table 1).All tree samples were collected from 1 or 2 m above ground, and grass samples were collected from approximately 2 inches above the root to minimize contact with soil.Once collected, the samples were rinsed three times with 10.1029/2023GB007796 3 of 10 Milli-Q water to remove any mineral dust that had accumulated on the plant matter.The cleaned samples were dried at 50°C in a standard laboratory oven.Pine needles, eucalyptus foliage, and grass herbage underwent additional cleaning (a rinse with a 50% methanol-water solution) to remove particulates that adhered to their waxy coating (Oliva & Raitio, 2003).After cleaning, a small portion (1-2 g) of each sample was analyzed for trace metal concentrations.
Clean foliage samples were burned outdoors on a secluded gravel lot in Gilbert, AZ, during three sessions (January, July, and September 2017).The fuel bed was constructed using cinder blocks and a ceramic floor tile.The floor tile was encased in aluminum foil to avoid Fe contamination.Cinder blocks delimited this arrangement to prevent the fire from spreading to the surrounding vegetation.Aerosol samples were collected using a high-volume sampler (Tisch Environmental), drawing in air at 1.13 m 3 min −1 .Aluminum ductwork (4″ diameter) was connected to the Teflon sampling stage downstream of an aluminum PM 2.5 impaction stage (Tisch Environmental TE 231) to isolate the fine particles (PM 2.5 or particles with aerodynamic diameter smaller than 2.5 μm).The Teflon collection stage was soaked in reagent-grade HCl (2.4M) overnight, rinsed, and dried before use.
At the beginning of each burn, small amounts of foliage were placed on the aluminum foil, sprayed with 1 mL of acetone, and ignited.To prevent damage to the PTFE sampler, the air temperature was kept below 300°C by  gradually adding small amounts of foliage to the fuel bed until the sample was consumed.Recent studies suggest that reed fires may be on the order of 300-500°C (Kurisu & Takahashi, 2019).Although the temperature in our experiments may be low compared to what would be expected in a crown forest fire, the sample was left until completely burned.Specifically, the foliage was kept aflame as long as possible, but the sample was left smoldering until cold.Higher temperatures would have damaged the Teflon collection stage or required the use of a metal collection stage which would have compromised our ability to measure Fe.Each burn lasted between 4 and  5 of 10 10 min.The cellulose filters were collected from the sampling stage using acid-washed tweezers and returned to their zipper bags.

Sample Processing and Analysis
Sample digestion was modified from previous studies and is briefly described here (Upadhyay et al., 2009).
Approximately 1 g of unburned foliage was placed into a PTFE-lined glass digester vessel (35 mL, Discover SP-D, CEM) with concentrated HNO 3 acid (10 mL) and Milli-Q water (5 mL).These vessels were microwaved at 240°C for 15 min.The volume of the resulting digest was transferred to a PTFE vial, reduced to <5 mL on a hot plate, transferred back to the digestion vessel using concentrated HNO 3 (10 mL), re-digested, and dried to a residue.Residues were subjected to a secondary hot plate digestion procedure as follows: concentrated HNO 3 and 30% trace metal grade H 2 O 2 (750:250 μL) for 2 hr; concentrated HNO 3 and HCl overnight (3:1 mL); concentrated HNO 3 and HF (4:1 mL) overnight; and concentrated HCl (5 mL) overnight.Between each step, sample containers were opened, and their volume was reduced to near-dryness.
One-quarter of each filter was digested via microwave (25 mL MARSXpress vessels, MARS 5, CEM).Filters were cut using zirconia ceramic blades (Specialty Blades, Inc.) on an acid-washed polypropylene cutting board and handled using acid-washed PTFE tweezers.Samples were placed into digestion vessels with concentrated HNO 3 (10 mL), concentrated trace metal grade HF (1 mL), and Milli-Q water (4 mL).Digestion vessels were heated over 30 min to 180°C.The volume of the resulting digest was reduced to <5 mL, transferred back into the digestion vessel with 10 mL HNO 3 , re-digested, and dried to a residue.The residue was treated with concentrated HNO 3 and 30% trace metal grade H 2 O 2 (500:250 μL) over a hot plate for 2 hr.
An aliquot of each digest was diluted using HNO 3 (0.32 M) and analyzed for 42 major and minor elements using a quadrupole inductively coupled plasma mass spectrometer (ICP-MS, iCAP Q, Thermo Scientific) at the W.M. Keck Foundation Laboratory for Environmental Biogeochemistry, Arizona State University.Polyatomic interferences such as 40 Ar 16 O + were removed using He as a collision gas.An internal standard containing Sc, Ge, Y, In, and Bi was used to correct instrumental drift throughout each run.The limit of detection for Fe measurements was <6 ng L −1 .The precision of Fe measurements was ±2% (2SD) based on repeated measurements of standard solutions interspersed within each run.

Blanks
Several types of blanks were collected during sampling and sample processing to account for all the sources of Fe in the experiment.A simulated sample consisting of only clean acid was processed alongside each sample set to quantify Fe in sample processing.The largest blank was found in the acid blank digested with unburned foliage (289 ng Fe).To measure the Fe contribution from the cellulose filter substrate, a one-quarter sample of six clean filters was digested and analyzed.The average amount of Fe on the filters was 1.7 ± 1.5 μg filter −1 .Particulate matter from the ambient air was collected on clean filters before, during (every three samples), and after each session for 5-10 min.These "field blanks" were processed as samples to determine the Fe contribution from ambient particulate matter.Our blank processes show that the filter blank had a higher Fe contribution than any of the acid blanks.Therefore, the Fe from the filter blank is the most important to consider when accounting for blank contamination.

Fe Concentrations of Unburned Plant Matter
Biomass burn experiments were conducted on 25 samples representing four biomes.The Fe concentrations of the unburned foliage samples, expressed relative to the total dry weight of the samples (g DW ), are shown in Figure 4. Pine, tropical, and eucalyptus foliage samples had concentrations between 10 and 125 µg Fe /g DW , consistent with previous findings (Hagen-Thorn & Stjernquist, 2005;Reimann et al., 2018).Differences among average Fe concentrations of the pine, tropical, and eucalyptus foliage were investigated using ANOVA at the 95% confidence level, and these three types of foliage are statistically indistinguishable (p = 0.86).Grass samples had an average Fe concentration of 250 ± 40 µg Fe /g DW , consistent with other measurements (Schlegel et al., 2016).The grass samples had high Fe content relative to the other plant material measured in this study (ANOVA, p < 0.5).It is important to note that the grass samples here were collected in Buffalo Park, Flagstaff, AZ, where the soils are rich in basalts due to the proximity of several dormant volcanoes.Because basalts weather easily and release Fe, the plants grown in basaltic soils may have high Fe and other trace metals than grasses grown elsewhere (Aiuppa et al., 2000;Burghelea et al., 2015).Furthermore, we use the results from the pine trees in Flagstaff to represent both temperate and boreal forests.Pine trees are a dominant class of vegetation in both regions and therefore can be used to approximate the Fe released from pine from the forest biome.However, collecting samples grown under the boreal and temperate biome conditions would provide better constraints on the Fe behavior under similar environmental conditions.Future work should target vegetation from each of these regions.

Fe Concentrations of Burned Plant Matter (PM 2.5 ) and Field Blanks
Fine particulate matter containing Fe was collected throughout the experiment (Figure 5).To ensure that the Fe concentrations on each filter were attributed to biomass, all samples, and field blanks were corrected for the average concentration of Fe in an acid-cleaned filter blank (e.g., 1.7 ± 1.5 μg filter −1 ).The samples and field blanks were also normalized by the particle collection time.
Many of the burn samples measured had Fe concentrations comparable to the field blanks.Figure 5 shows the concentration of Fe that was collected on the filter after a blank correction.The higher field blanks may be due to changing wind conditions during sampling.Additional mineral dust aerosols might have entered the air sampler if the wind speed or direction shifted during sampling.Because mineral dust aerosols have an order of magnitude more Fe than some of the burn aerosols (e.g., pine and eucalyptus leaves), even a small increase in mineral dust could increase the amount of Fe collected on the field blank (Duce & Tindale, 1991;Mahowald et al., 2005).However, the field blanks only had an average of 1.5 μg of Fe on the filter, which is comparable to the Fe in the ponderosa pine samples and eucalyptus samples.Despite the similarity of the Fe concentration in blanks and the samples, the amount of Fe accumulating is low.Thus, any blank contamination will not impact the overall mass  balance.Additionally, many inferences can be drawn from this data set despite high field blanks relative to some aerosol samples.Specifically, these measurements can be used to place constraints on the percentage of Fe that was aerosolized during the experiment and allow us to calculate a global estimate of the Fe flux from biomass burning.
Although some of the samples and blanks may have been influenced by mineral dust in the air during the burn, burned pine needle aerosol samples may provide insight into the relative impact of ambient aerosols.We can treat the pine needle foliage samples as replicate measurements.We find that the relative standard error between the pine samples is 29%.Replicate measurements of similar biomass burning experiments conducted by the U.S. Forest Service's Fire Sciences Laboratory (FSL) at Missoula report a 16% standard error (Yokelson et al., 1996).Unlike the FSL experiments, our experiments were conducted outside where air currents are less predictable.Therefore, we expect a slightly larger standard error as we cannot preclude an Fe source from ambient air.
The percentage of Fe aerosolized from each sample can be estimated using Equation 1: where Fe aerosolized, fine is the filter-blank-corrected amount of Fe in each sample in μg Fe filter −1 min −1 , Fe fieldblank is the average field blank in μg Fe filter −1 min −1 , and Fe foliage is the amount of Fe in the unburned foliage in μg Fe .
None of the samples exceeded 2% Fe aerosolized (Figure 6).Each percentage of aerosolized Fe was weighed to calculate a weighted average.To calculate the weighted average, the samples within a foliage type were taken as replicates of the content of that biome.Several samples had Fe concentrations that were indistinguishable from the field blanks.These samples were forced to have 0% Fe aerosolization to avoid negative percentages.Thus, these weighted averages are 0.00% ± 0.01%, 0.4% ± 0.1%, 0.00% ± 0.04%, and 0.06% ± 0.02% for pine, tropical, eucalyptus and grass samples, respectively.

Estimate of Global Fe Flux From Biomass Burning to Ocean
To approximate a global Fe flux of biomass burning to the oceans, we use our aerosolized estimates in combination with estimates of biomass burning and fuel density.The Fe flux to the atmosphere in small particulate matter was calculated according to the following Equation 2: where Fe aerosol, fine is the percentage of Fe aerosolized in Gg Fe yr −1 , Fe foliage is the Fe concentration of unburned foliage in ppm, density fuel is the estimated foliage fuel density of each biome in Mg ha −1 , and area burned is the annual area of each biome subject to biomass burning in Mha (Randerson et al., 2012).The values used here are shown in Table 2.We assume that total biomass density is roughly equal to the foliage density for regions with little woody input, such as the savannah, croplands, and grasslands.For the forests, we employ a foliage density of 5 Mg ha −1 (Zhang & Kondragunta, 2006).
Outliers were excluded from the data set using a Q test (p < 0.5).One Pinus ponderosa sample and the Pachira aquatica were excluded from the unburned data set, and Bouteloua gracilis was excluded from the aerosolized Combining Ito (2015) and our estimations, the fine fraction derived from burning foliage accounts for only 2%-9% of Fe from biomass-burning aerosols.The rest of the Fe flux can be attributed to the Fe in the coarse fraction and the Fe released from the soil.Our study was unable to quantify the amount of Fe that is released directly in the coarse fraction during burning events.However, previous studies suggest that the ratio of fine to coarse particles is approximately 1:4 (Ito, 2013;Luo et al., 2008;Wang et al., 2015).If we apply this ratio, then the coarse fraction accounts for 8%-36% of the emitted Fe.Thus, we suggest that lofted soil particles comprise the majority of Fe released during biomass burning events (55%-90% of Fe released).
Although the contribution of Fe from fine pyrogenic particles is small, its importance may be large because of its increased solubility (Ito et al., 2021).Although a significant portion of bioavailable Fe delivered to the oceans is derived from dust in arid and semiarid regions (Guieu et al., 2005), the smaller size of fine particles created by burning lends itself to longer residence times and increased atmospheric processing (Jickells et al., 2005).More work is required to determine the solubility of particles after they have undergone a burning event and subsequent atmospheric processing.Furthermore, our data do not allow us to include a discussion of Fe speciation or a discussion of oxidation state, which is important in determining how pyrogenic Fe reaches marine environments.Future work is needed.

Conclusions
This study provides a crucial first-order estimate of global plant-derived Fe from biomass burning.By using controlled burn experiments on representative plant materials from an array of different biomes, we estimate that the percentage of biomass Fe in fine particles (PM2.5)aerosolized during burning events is <2%.Furthermore, our results show that the influx of plant-derived Fe from the fine particles is small relative to Fe from soil-lofted particles.We estimate that the fine fraction of Fe from burning foliage constitutes less than 9% of the total Fe released during a burn.
However, while we find that the overall contribution of fine particulate Fe is relatively small compared to soil-lofted particles, their increased solubility may make them an important Fe source to marine ecosystems.Because of differences in solubility, it is important to disentangle the various types of particulate Fe that are released during burning events.While we successfully measured the fine fraction Fe flux, future work is needed to better constrain the coarse particle fraction.Additionally, more studies are required to examine how the atmospheric processing of pyrogenic particles affects their solubility.Until such studies are carried out, it will be challenging to truly understand how Fe from biomass burning emissions can impact ocean biogeochemistry.

Figure 1 .
Figure1.Burn experiment setup.The foliage was burned within the fuel bed (detailed in Figure2).The resulting combustion particles were drawn through the aluminum ductwork via the volumetric flow controller (VFC) sampler and impacted clean cellulose filters at the sampling stage (Figure3).Equipment was assembled upon a tarp to decrease soil entrainment during experiments.

Figure 2 .
Figure 2. Contained fuel bed.Aluminum foil-covered concrete blocks surrounded each burn.The fuel bed consisted of an aluminum foil-covered ceramic floor tile on top of a 50 × 50 × 0.3 cm aluminum sheet and aluminum foil-covered ceramic blocks to prevent tarp melting.The material for each burn was placed on top of a fresh piece of aluminum foil to prevent cross-contamination between samples and facilitate post-burn ash collection.

Figure 3 .
Figure3.Sampling stage.To prevent contact between the filter and the steel mesh of the volumetric flow controller sampler, the sampling stage was constructed with a polytetrafluoroethylene (PTFE, Teflon) stage and impactor.PTFE and other plastic components were soaked in 20% trace metal grade HCl between each burn experiment.The aluminum impactor was cleaned before and after each burn using Kimwipes.

Figure 5 .
Figure 5. Concentrations of Fe in the filter blank-corrected samples.The average measured concentration of Fe in clean filters was subtracted from the measured amount of Fe in each sample.

Figure 4 .
Figure 4. Fe concentrations in unburned foliage (μg Fe ) normalized to the dry weight (g DW ) of material analyzed.

Figure 6 .
Figure 6.Percentage of Fe aerosolized in PM 2.5 from the unburned foliage during biomass burning experiments.The average field blank has been subtracted from each sample.Error bars on the data points represent propagated errors from the field blanks.

Table 1 Sample
Details Including Species, Total Dry Weight, and Sampling Location 10.1029/2023GB007796