Chemical characterization of pellet stove ashes: Elemental fractionation of ashes results in volatile particles with high concentrations of cadmium and lead

Abstract

Use of wood-based biomass for home heating has increased as a more sustainable and economical alternative to fossil fuel heating sources. However, concerns remain regarding particulate emissions and potential human health effects. Pellet stoves are very efficient combustion sources but still emit high particle numbers of nanoparticles into the environment and generate ash within the stove that could be an exposure source during cleaning. In this study we chemically characterize pellet ash collected from zones (upper and lower) from three stoves and determine the chemical composition of room air during cleaning events. Ash was acid digested using reverse aqua regia and microwave digestion and analyzed by ICP-MS. Indoor air was sampled using impingers before and during stove cleaning to assess ultrafine particulate release and exposure. Within the pellet stove different fractions of ash could be identified by color and macro-scale morphology. Ash fractions collected lower in the stove were predominantly calcium-based, 25% by weight, and high in refractory trace elements, Fe, Cr, Mn and Ni. In contrast, ash collected higher in the stove was predominantly K and S based (25 and 5% respectively) and had high concentrations of volatile elements Zn (0.4%), Cd and Pb. Maximum Cd and Pb ash concentrations were 274 mg/kg and 807 mg/kg respectively corresponding to enrichment factors of 1581 and 870 respectively compared to the respective pellets. Electron microscopy of the particles suggested that ash particles were agglomerates of smaller primary particles and EDS identified areas of high K and S concentration on the particles. Impinger studies of room air during cleaning showed an increase in K, S, Zn, Cd and Pb suggesting a mobilization of upper ash particles into room air upon stove cleaning.

1. Introduction

Many people consider using biomass as a fuel source to be a more renewable practice than using fossil fuels and, as such, home heating with wood has increased in popularity in the 21^st century[1], although the assertion that burning biomass as a fuel is truly a renewable practice is increasingly disputed [2]. Home heating with wood fireplaces or pellet stoves has long been common in rural areas and is particularly common Northeastern US states[3] compared to the rest of the US. In 2014, 2.1% of US households reported using wood as their main heating source and an additional 7.7% as a secondary source[4]. Wood and pellets are considered lower cost heating alternatives than oil or kerosene but concerns remain regarding particulate pollution of indoor and outdoor air from wood burning[5–8]. The hazards of particulate exposure are associated with particle size; particles < 10 um (PM₁₀) are respirable, smaller particles (< 2.5 um, PM_2.5) can penetrate deeply in the lungs and ultrafine particles (≤ 0.1 um, nanoparticles) are thought to be most harmful in this regard[9–11]. Commercial/residential combustion sources have been estimated to cause 10,000–40,000 premature deaths annually in the US, largely clustered in the central and northern east coast regions[12, 13].

In 2015 the US EPA released the EPA Certified Wood Stove Database and began to actively encourage home owners to replace older non-compliant stoves with certified stoves which are stated as more energy efficient and less polluting, although the latter assertion has been questioned [14, 15] }. In addition, pellet stoves have been considered as a lower pollutant energy source[16], as they are highly efficient in terms of combustion[17], which reduces particle pollution, in large part because more organic carbon is combusted.

A particular exposure scenario of concern for wood stoves is during restocking the wood, which can mobilize ash particles into the air[5], whereas pellet stoves are considered by many to be of less of a concern in this regard. However, pellet stoves require regular interior cleaning of ash as this builds up during the heating season. During cleaning of pellet stoves ash particles can be mobilized into the air and this could be a concern for human exposure. While prior work shows that pellet stoves predominantly produce sub-micron particles in the combustion gas stream[18, 19], there is little information on the physical and chemical characteristics of pellet ash remaining in the stove and released indoors during cleaning. Measuring particle composition is time-consuming and expensive compared to measuring particle size and for the most part cannot be continuously monitored. Nevertheless, it is logical to hypothesize that the toxicity of particle exposure can be related to the chemical composition of the particle as well as particle size[20, 21].

Biomass ashes are primarily inorganic, and their composition can be broadly characterized by end-members composed of ‘non-volatile’ elements dominated by Ca, P, and Mg, or of volatile elements dominated by K, S, Cl and Zn[22, 23]. The K-based volatile ash particles are the main constituents of particulate pollution emissions from wood combustion sources[18, 22, 24–27] and therefore pose the major human exposure route and health effects. Compared to characterization of particle size and number concentrations from pellet stoves there have been relatively few studies focused on comprehensive characterization of elemental concentrations of ash[28] and the variation in the concentrations of toxic trace elements between ‘bottom’ ashes and more volatile ash. The objective of this study was to characterize the elemental composition and particle size distribution of ash collected from different zones within pellet stoves and to characterize the indoor atmosphere during stove cleaning.

2. Methods

Pellet ash samples were sampled on three different pellet stoves (A,B,C) in use at three different residential households in Hanover, NH, USA. The make and model of stoves A, B and C are a Piazzetta Sabrina, an Enviro Empress FPI (Fireplace insert), and a Harmon Mark XXV, respectively. The design of each stove was somewhat different, but in general ashes were collected from the firebox and/or below the firebox (lower), or from the sidewalls and from the back and top of the stove (upper) (Figure 1) for each stove. Ash was collected from each area by brushing the ash with a toothbrush or cloth onto a piece of A4 paper and then stored in a 50 ml polypropylene centrifuge tube at room temperature.

FIGURE 1:

Interiors of stoves A, B and C with stove door open. Stove A has back plate removed, stove B shows individual areas of ash sampled, a sampling procedure also followed for stoves A and C.

At one residence (Stove A) the indoor air was sampled during stove cleaning using an impinger filled with 30 mls of deionized water (Element Qpod, Millipore) through which 4L per minute of room air was bubbled by attaching a pump to the side arm of the impinger. Two impinger samples for room air ‘blanks’ were first collected with no room or stove activity during collection, the DI water from each impinger was decanted into 50 ml tubes, the impingers were refilled with DI water and impinger samples were collected during stove cleaning event. This cleaning activity was sampled on two different occasions, the duration of the first sampling was 30 minutes for both blank and cleaning event. For the second sampling the duration was extended to two hours for both blank and cleaning event, recognizing that the room air would remain high in particles after the cleaning event and to try and capture this by sampling longer.

2.1. Pellet Characterization

A Representative sample of pellets was supplied by each household. For household C, two different brands of pellets were in use around the time of ash collection and were supplied for analysis, and the average pellet concentration value was used for calculation of elemental enrichment factors for ash C The pellets from each household were homogenized, acid digested and analysed for a suite of elements are described below for ash. Percent moisture was also determined for each pellet type by gravimetric methods after heating for 4 hrs at 100 C.

2.1.1. Pellet and Ash Digestion

0.25 g of ash samples were acid digested using a 3:1 HNO₃/HCl trace metal grade acid mix (reverse aqua regia) closed vessel microwave digest with a 20 min ramp time to 200 C and 15 minute hold time. The resulting digestate was rinsed into a 50 ml tube using deionized water with additional rinses of the digestion vessel into the tube up to a final volume of 50 ml. NIST SRM 2711, Montana soil, moderately elevated trace elements, was used for quality control.

2.1.2. Electron Microscopy

Ash particles from upper ash samples were sprinkled on tape and imaged using a Thermo Scientific Helios SEM with major elemental analysis up to 5 KeV collected by EDS using an Oxford Instruments UltimMax 100 at the Dartmouth College Electron Microscope Facility.

2.1.3. ICP-MS analysis

Digested and aqueous samples were analysed for major and trace metals by triple quadrupole (QQQ) ICP-MS (Agilent 8900, Wilmington, DE). Triple quadrupole instruments utilize fully scanning quadrupoles before and after a collision cell, thereby controlling the analytes entering and exiting the collision cell and eliminating potential polyatomic or isobaric interferences when using reaction cell approaches. Their performance (sensitivity, background, selectivity) is generally considered to be superior to single quadrupole ICP-MS instruments[29]. The instrument was operated in He and O₂ gas modes; most analytes were measured in He mode except P, S, V, As and Se which were measured in O₂ mode by mass shift. Calibration was performed with NIST-traceable multi-element standards (blank and six calibration standards of increasing concentration) and the calibration checked by a second source NIST-traceable multi-element solution run after calibration and every 10 samples and a USGS water sample run after calibration.

2.1.4. Data Analysis

All statistical analysis was performed in JMP 16 (Cary, NC). Basic data handling and preparation of figures was performed in Microsoft excel. To estimate total recovery based on the elemental concentrations the ‘total’ inorganic constituents of the ash were approximated by apportioning all S to K₂SO₄ and all P to CaHPO₄ and remaining K and Ca and all other measured elements to their respective oxides. This approach makes many assumptions, our analysis omits Si, Ti and Cl which could form significant solid phases and we do not have an independent measurement of residual organic matter (which, based on other studies we expect is < 3%). Nevertheless the assumptions are reasonable for a semi-quantitative approach, are consistently applied across the three stoves and therefore allow for a relative comparison between the stoves and lower and upper ash. The pH of these ashes were > 12 and this supports our supposition that metal oxides or hydroxides are a major solid phase for the alkaline and alkaline earth metals.

3. Results

Ash was collected from three stoves in 2019–2020. Ash was denoted as either lower or upper depending on where, in relation to the firebox, it was collected from, with ash from in or below the firebox being designated lower and anything above the firebox designated upper. A summary of the major and trace element composition of the individual ash samples taken from the stoves is shown in Figure 2 and a comprehensive presentation of concentrations for 44 elements in the pellets and their respective ashes are given in supplemental table 1 and 2. The % moisture for the pellets was low, being 3.8, 3.0, 2.9, 2.4 % for A, B, C1 and C2 respectively.

Figure 2:

Major and trace elements in ash fractions of three pellet stoves (A, B, C). Multiple locations were sampled within the ‘Lower’ and ‘Upper’ designated areas for each stove.

For the major elements in Figure 2 (left panel), ash collected from the lower potions of the stoves was higher in Ca, Mg, P and Mn while ash from the upper regions of the stoves was higher in K, S and Zn. For the trace elements the lower ash had higher concentrations of V, Co, and Ni, while upper ash was higher in Li, Cu, As, Se, Cd, and Pb. Principal components analysis for the data in table 1 based on covariance led to one component explaining 97% of the variance, and the analytes with positive covariance were those that were enriched in upper ash (Li, Na, S, K, Cu, Zn, As, Se, Rb, Mo, Ag, Cd, Sb, Cs, Tl, Pb) whereas those with negative covariance were those found enriched in Lower ash (Be, Mg, Al, P, V, Cr, Mn, Fe, Co, Ni, Sr, Ba, Ce, rare earth elements, Th, U). Digestion recovery based on the sum of the calculated solid phases ranged from 58% - 81% and was significantly higher for the upper ashes compared to the lower ashes. Lead and Cd showed a large concentration increase in upper ash with maximum concentration of 807 and 274 mg/kg enriched 44 and 84 times over the mean lower ash concentrations. Lead and Cd were highly correlated with S, and Pb was also highly correlated to K and inversely correlated with Ca (Figure 3).

Figure 3:

Lead and Cd are enriched in K/S-rich ash and depleted in Ca-rich ash

Enrichment factors of element concentrations in lower and upper ashes compared to the original pellets were calculated by ratioing the concentrations of each element in the ash to the pellet concentration. Average lower and upper concentrations for ash were used for each of the stoves. Sulfur, K, Cd, Pb and Zn, i.e volatile elements, are highly elevated in ash from upper regions of the stove with maximum EF of 570, 391, 1581, 945, 643 respectively, and depleted in lower regions with maximum EFs of <32, 126, 45, 79, 40. The converse is observed for non-volatile elements Al, Fe and Mn, which are generally higher in ash collected from lower in the stove and lower in ash from upper regions (Figure 4).

Figure 4:

Enrichment of trace elements in ash from three different pellet stoves. Enrichment is calculated as concentration of the element in the ash relative to the pellet concentration.

Electron microscopy and EDS of ash particles from one upper ash sample showed spherical and irregular shaped particles that appear to be agglomerates (< 25um - 250 um) of smaller particles, with zones of high K and S concentrations (Figure 5). Although the incident energy of the EDS (5 KeV) was not sufficient to detect higher X-ray energy elements, it can be surmised from the relationships revealed in the total data that these high K and S zones were also high in Cd and Pb.

Figure 5:

Electron microscopy-EDS of upper ash particles shows them to be agglomerates with enriched zones of K and S

To assess whether indoor air was impacted by airborne particulate contaminants during stove cleaning, air was sampled using an impinger which can trap particles in a water trap through which indoor air is bubbled[30–32]. The concentrations of relevant elements in the DI water trap are shown in Figure 6. The average elemental concentration enrichment in DI water during stove cleaning relative to the room blank air prior to cleaning were in the order Cd > K > S > Mn > Zn > Mg > Ca > Pb >> Ni, Na, Cr, Fe, Co, Cu. Stove cleaning concentrations for Mn, Mg, and Ca were skewed higher due to one high replicate and only Cd, K, S, Zn and Pb exhibited significantly higher concentrations during stove cleaning compared with equivalent sampling prior to stove cleaning (here termed room blank).

Figure 6:

Impinger water concentrations (n= 4) collected from room air at 4L/min for 90 minutes showing significant increases in K, S, Cd and Pb during stove cleaning.

4. Discussion

The major findings of this study are the extremely high enrichment of the toxic elements Cd and Pb in ashes collected from the upper regions of all three pellet stoves and the observation that these particles appear to be released to the room atmosphere during stove cleaning. Concentrations of these elements in the raw pellets were < 1 mg/kg for each element but reached >200 mg/kg for Cd and > 800 mg/kg for Pb in upper ash, equivalent to maximum enrichment factors of ca. 900 – 1600 when ratioed to the concentration in the pellets. The combustion efficiency of pellet stoves is high, with residual ash contents generally ca. < 3%, and with values as low as 0.3% being reported from laboratory studies[33]. Hence, with no elemental fractionation of the ash, just concentration from mass loss due to combustion of organic matter, enrichment factors of 30 - > 300 could be expected in the residual ash for combustion efficiencies of 97 – 99.3%. Our experimental set up precluded any quantitative measurement of combustion efficiency based on weight of pellets combusted and the residual ash produced, however we can estimate an average combustion efficiency using the Cu concentrations of the pellets and ash. Copper concentrations in the four different pellet stocks were quite similar (1.32 ± 0.08 mg/kg, Table 1) and although slightly higher in concentration in the volatile upper ashes, the ash concentrations were also relatively consistent (lower: 184 ± 23, upper: 280 + 56 mg/kg), this leads to average enrichment factors for Cu of 139 and 211 for upper and lower ashes or an average value for all ash of 175 which suggests a combustion efficiency of 99.6%. Therefore, elements enriched in either the upper or lower fraction should have EFs significantly greater than 175, which also would require that their EF for the other fraction would be significantly less than 175. Enrichment of Cd and Pb is clear with average EF in upper ashes of 1187 and 723 respectively and EFs of only 29 for each element in the lower ash and strongly indicates volatilization of these elements in the stove and subsequent condensation onto particles at higher locations in the stove. Potassium, Zn, S and Ag were also significantly enriched (average EFs of 255, 358, 500, 518 respectively) in the upper ashes and ‘depleted’ in the lower ashes (103, 26, 38, 27, respectively). The depletion/enrichment of the non-volatile elements is less striking, presumably because the ash particles are a continuum and smaller lower ash particles seed the upper ash particles. Average enrichment factors for Al, Mn Fe and Co in lower ash are 345, 183, 259, 239, compared to upper ash factors of 165, 99, 167, 162 respectively.

The range of Cd concentrations of the four pellet stocks used in this study (0.09 – 0.209 mg/kg) is slightly higher than the 0.00009 – 0.14 mg/kg range reported in a comprehensive study of the chemical composition of 132 wood pellet samples from the NE USA[33] while the Pb range of 0.552 – 0.963 mg/kg in this study is within the 0.04 – 11 mg/kg range reported for NE USA. A European study drawing on 249 pellet samples data from a variety of databases reported Cd range of 0.14 – 43 mg/kg, and median of 0.2 mg/kg and Pb range of <2 – 57, and median 2 mg/kg[22]. Hence the pellet values of Cd and Pb herein are in line with US and European pellets and do not suggest contamination of the pellets themselves. However, the ash concentrations of Cd and Pb reported in this study are, to our knowledge, the highest reported for wood/pellet ash samples albeit that other studies do not sub-sample ash from different parts of the stove. Lead concentrations ranging from 238–637 mg/kg[34] and 22–640 mg/kg (< 100 um fraction)[35] have been reported for wood ashes although the concentrations of Cd in the ash reported in these two studies are 0.5–1.1 and 3.4 −11.2 mg/kg respectively, much lower than results reported here. One study, focused on the use of ash as a soil amendment, did report higher values for Cd in pellet ash with Cd and Pb values of 41 and 44 mg/kg respectively[36].

Cadmium and Pb concentrations in the different zones/types of ash sampled in this study are highly correlated to major volatile ash elements K and S, which are known to also be major constituents of the sub-micron particulate contamination that is released into indoor and outdoor air[37–40]. Potassium is the main volatile ash forming constituent[41] and it had been suggested that identification of low K wood, or exclusion of high K wood to keep K concentration to < 500 mg/kg in pellets could be used as a measure to reduce PM formation and contamination from pellet stoves[22]. Median K values for the 132 NE US wood pellet samples were 709 mg/kg with a range 167 – 9833 mg/kg[33], while the European study of 249 pellet samples had a K median of 540 mg/kg and range of <17 – 1160[22]. The four pellet brands in this study ranged from 741 – 960 mg/kg, agreeing with the NE US study and suggesting that pellet stoves in NE US may generate excess PM due to the high K of regional pellet stocks. Potassium chlorides, sulfates and carbonates have been identified as the main inorganic particles in PM from pellet ash[38]. Because of the high enrichment of Pd and Cd in the upper ashes and the strong correlations with K and S we hypothesize that particulate Pb and Cd would also be released to indoor air during stove cleaning, and to a lesser extent during normal stove operation. The results of our impinger sampling show increases in K, Cd, Pb, S and Zn during stove cleaning which suggests mobilization of particles of these ashes into the room air. The multi-element fingerprint of the impinger trap is very similar to that of the upper ash and we hypothesize that the upper ash particles we measured to be high in K, S, Cd and Pb are the source of nanoparticles in the room air. Further research in characterizing these particles is recommended here. A laboratory-based study using an entrained flow reactor, varying temperature, O₂ and SO₂ levels and trapping sub-micron particles using a cascade impactor found Pb concentrations associated with the ultrafine particulate fraction to be ca 738 - > 5000 mg/kg in the sub-micron fraction under the various test conditions and burning orujillo, an olive oil solid bioproduct as fuel[42]. In this case the Pb concentration of the fuel was high for biomass, being 18.4 mg/kg, nevertheless this equates to EFs up to ca 300 for Pb and illustrates that Pb is highly concentrated in the sub-micron ultrafine particulates. Another recent study reporting flue gas particulate concentrations for 5 pellet ashes found levels of Pb in these particles to be ca.1000–5000 mg/kg in the ultrafine fraction from burning of wood pellets with Pb concentrations of 0.2 – 3.1 mg/kg[41]. Both of these studies support the findings reported herein.

A comparison between sub-micron particle number emissions from residential heating systems (natural gas and liquid petroleum gas boilers and pellet stoves) found that the highest concentrations were measured for pellet stoves[19], the particle mass emitted by wood and pellet stoves is dominated by fine particles (< 1um) and for pellet stoves is dominated by inorganic particles of K₂SO₄ and KCl and Pb, Cd, and Zn had maximum concentrations in this fine fraction[18]. Furthermore a study of indoor and personal exposure from wood burning stoves (PM2.5) found wood-smoke particles significantly increased K and Zn concentrations and suggested that Pb could be marker of indoor exposure to wood smoke[43]. The observations from these previous studies are consistent with, the findings of this current study which exemplify that particulate contamination from pellet stoves is comprised of inorganic elements, predominantly K and S based and with high concentrations of Pb, Cd and Zn.

While the major element trends of the upper and lower ash were consistent between the stoves, some individual stove differences were observed. The upper ashes of Stove B had more ‘lower’ ash character than stoves A and C, as can be evidenced by the higher Ca component of upper ash in Stove B. The stoves are produced by different manufacturers and are of different designs, the reduction in EFs of the volatile elements in upper ash for stove B could be due to lower combustion temperatures or differences in air/oxygen supply during combustion. Finally, regarding technological critical elements, lithium was enriched in the upper ash for all stoves, and up to 100 – 200 mg/kg for stove A, whereas the light rare earth elements and cobalt were enriched in the lower ash. While this is an interesting observation it is unlikely that concentrations are high enough to make recovery of these metals from ash commercially viable.

5. Conclusions

While this is a limited study of three pellet stove ashes and does not account for other individual sources to trace elements in the three households, the strengths are a comprehensive compositional analysis and impinger sampling of room air. The results of this study and those of other biomass burning studies indicate that particulate matter from wood and wood products has a distinct elemental fingerprint, is present in sub-micron particles, which can penetrate deep in the lung and contain high concentrations of Cd and Pb. In addition to being an indoor exposure source, it is likely that these particles are vented to the outdoor environment through the stove exhaust. Other studies have focused on combustion of contaminated wood products, and undoubtably a higher contaminant concentration in the source wood will result in higher concentrations of volatile elements in the ash. However, we show in this study that high concentrations of Cd and Pb in ash particles are produced even from low pellet fuel concentrations through the combustion efficiency of modern stoves and the intrinsic volatility of these elements. We suggest that modern efficient wood burning stoves would exhibit the same elemental fractionation in ash elemental composition with a concentration of Cd and Pb in the fine particles. These particles could have additional health effects due to their toxic trace element composition. Practically, we advise owners of pellet stoves to wear KN95 face masks while cleaning the stove and handling the ash. Use of a room HEPA filtration unit can assist in removing particles from indoor air. We do not advise using pellet ashes as liming agents for home gardens due to the high levels of Cd and Pb and potential for uptake by plants and surface contamination of root vegetables.

Abbreviations:

ICP-MS

Inductively coupled plasma mass spectrometry

SEM

scanning electron microscopy

EDS

energy dispersive spectroscopy

NIST

National Institute of Standards and Technology (US)

SRM

standard reference material

USGS

United States Geological Survey

EF

enrichment factor