Bioremediation of engine-oil contaminated soil using local residual organic matter

Soil preparation

This experiment was conducted on a certified treatment platform for contaminated soils owned by SolNeuf Inc., and operated by AKIFER Inc., located in Neuville, QC, Canada (46°44′5.0352″N, 71°41′2.634″W). The soil used was sandy (98.4% ± 0.3% sand, 2.5% ± 0.2% gravel, 0.9% ± 0.1% silt and loam), with a pH of 6.4 (1:1, soil:H₂O). It contained less than 0.3% organic matter (combustion) and 34.4 mg kg⁻¹ phosphorus (P). The soil was free of PHC as measured by gas chromatography with flame ionization detector (GC-FID) according to the method MA. 400—HYD. 1.1 (Centre d’expertise en analyse environnementale du Québec (CEAEQ), 2016). In order to control the concentration and homogeneity of the contamination, the initial soil was artificially contaminated with unused engine oil to 7,500 mg kg⁻¹ of PHC (C₁₀–C₅₀) and thoroughly mixed for 1h by excavator. The oil was RUBIA LD 10W30, manufactured by Total for diesel motors. The PHC concentrations were followed from June to November 2015. The target for remediation was the Government of Quebec’s “C” Criteria (less than 3,500 mg kg⁻¹ in the C₁₀–C₅₀ range) (Beaulieu, 2016).

RFM soil amendments

For the RFM materials, we defined as “local” the ones that originated from a distance of 100 km or less of the soil treatment platform. This action was guided by a circular economy framework and the Interstate Technology & Regulatory Council (ITRC) (2011) Regulatory Guidance for Green and Sustainable Remediation to minimize the project’s carbon footprint by reducing fuel consumption associated with transport. A recycling and triage center located just 600 m from the treatment platform was the first location where RFM were sought out. RCW of mixed origin (unspecified mix of deciduous trees and conifers, excluding Thuja sp.) was chosen. RCW differs from regular wood chips because it originates from branches less than seven cm in diameter and not whole trunks (stem wood) (Lemieux, 1986). The second closest available RFM was excess MANR from stables eight km away (homogeneous mix of fresh to 8 months old manure). The last RFM was BSG. Initially intended to be sourced 12 km away, the BSG used for the project had to be sourced 92 km away because of time constraints. The spent grain used in this project was 4 days old and had been stored outdoors in barrels.

Experimental design

The four treatments used were: 1—Ramial Chipped Wood (RCW-F), 2—Brewer’s Spent Grain (BSG-F), 3—Horse manure (MANR-F), and 4—Fertilizer alone (FERT) (88.6% calcium mononitrate (27 N:2.7 Mg:4.6 Ca) and 11.4% diammonium phosphate (18 N:46 P:0 K) to achieve a C:N:P ratio of 100:2.5:0.5). This fertilizer dosage is the usual treatment at this treatment platform, and it was also added as a nutrient baseline to all other treatments (“F”). Due to space and logistical constraints, no bins were tested without fertilizer. The organic amendments were added at a ratio of 30% by volume for each local organic amendment and were mixed for 10 min by excavator, followed by 10 min of hand-shovel mixing (see Fig S1 for design). For each treatment, three concrete cylinders (0.76 m³) were placed outdoor and filled with 760 L of the resulting mixtures of soil and amendments. The 12 cylinders were randomly distributed in a straight-line oriented southwest-northeast. Following the province of Quebec’s regulations, the tops of the bins were covered with a thick plastic and a pulled-air system was connected to a perforated 2″ PVC pipe placed 48 cm from the bottom of each bin (10% below the center). The air was drawn at a rate of 1.5 ± 0.26 m³ h⁻¹ and passed through a biofilter prior to release to the atmosphere. The rate was set to be proportional to the airflow in large biopiles on this site. Finally, all water runoff was collected and sent for treatment at a registered facility. The targeted soil moisture concentration was set at 70% of field capacity. When the moisture levels dropped significantly below that target, watering events took place (n = 3).

Sampling

The concentrations of C₁₀–C₅₀ PHC in the soil were monitored three times over 5 months, on June 10, August 31 and November 2, 2015. On the first and last sampling dates, the C₁₀–C₅₀ PHC fractions were split into six smaller fractions (C₁₀–C₁₆, C₁₆–C₂₂, C₂₂–C₂₈, C₂₈–C₃₄, C₃₄–C₄₀, and C₄₀–C₅₀) to quantify if there was preferential fraction degradation. At each sampling event two new holes were drilled with a manual auger (five cm diameter). Samples were taken from 30 and 60 cm depths in the first hole, and from 60 to 90 cm depths in the second hole. These four samples were homogenized and sub-sampled for all analyses. Nutrient measurements were conducted from the first and last sampling events described above. For moisture readings, the soil was sampled eight times and was measured gravimetrically (oven dry). Gas measurements were taken from a tube linked to a perforated PVC pipe capped on both ends and placed in the center of each bin. Carbon dioxide (CO₂), oxygen (O₂), and volatile organic compounds (VOCs) were measured with a portable gas detector (Eagle model; RKI Instruments, Union City, USA). The tube’s end was equipped with a clamp, which was released to measure the gases present in the bins. Temperature loggers (Levelogger Gold model 3001; Solinst, Georgetown, Canada) were placed inside the PVC pipes used for gas monitoring in each experimental unit. Temperature was recorded every hour throughout the experiment. The loggers were calibrated at the beginning of the experiment and the deviation between loggers did not exceed 0.3 °C (0.14 ± 0.03 °C (air) and 0.08 ± 0.03 °C (water)).

Analytical methods

The PHC fractions were measured by GCFID according to the MA. 400—HYD. 1.1 method (Centre d’expertise en analyse environnementale du Québec (CEAEQ), 2016). Soil samples were first dried with acetone (CAS no 67-64-1) and then extracted with hexane (CAS no 110-54-3) using a “paint mixer” type extraction system. Subsequently, silica gel (60–200 mesh grade 62 (CAS no 112926-00-8), SiO₂) was added to the extract to adsorb the polar substances. Finally, the supernatant hexane was analyzed by GC-FID. The concentration of hydrocarbons present in the sample was determined by comparing the total area of all peaks from n-C₁₀ to n-C₅₀ with the surfaces of the standards used to establish the calibration curve under the same assay conditions (diesel standard solution altered to 50% at 5,000 μg mL⁻¹ (Diesel fuel No. 2; Restek, Bellefonte, PA, USA). The methodological limit of quantification was 100 mg kg⁻¹ and PHC recovery rates were all between 82% and 94%.

The determination of ammoniacal nitrogen was a two-step process conducted according to method MA 300-N 2.0 (Centre d’expertise en analyse environnementale du Québec (CEAEQ), 2014a). The first step was an extraction in the presence of potassium chloride (CAS no 7747-40-7). Secondly, the ammonium ion reacted with sodium salicylate (CAS #54-21-7), nitroferricyanide (CAS no 13755-38-9) and dichloroisocyanuric acid (CAS no 2893-78-9) to form a blue alkaline complex with an absorbance at 660 nm, which is proportional to the concentration of ammoniacal nitrogen. Ammoniacal nitrogen recovery rates were between 95% and 96%. Phosphorus concentrations were determined according to method MA. 300—NTPT 2.0 (Centre d’expertise en analyse environnementale du Québec (CEAEQ), 2014b). Nitrate and nitrite ions were analyzed in accordance with method MA. 300—Ions 1.3 (Centre d’expertise en analyse environnementale du Québec (CEAEQ), 2014c). The soil was mixed with water in order to dissolve the extractable anions. For leached nitrates and nitrites, the extraction was done with the leaching buffer as specified in the Hazardous Materials Regulations and described in method MA. 100—Lix.com. 1.1 (Centre d’expertise en analyse environnementale du Québec (CEAEQ), 2012). Subsequently, anions are were separated by an ion exchange column using the following eluent solution: 0.0027M sodium carbonate (CAS no 497-19-8) and 0.0003M sodium bicarbonate (CAS no 144-55-8)). The retention time differs for each anion, which makes it possible to identify and dose them. Nitrates and nitrites were measured using a conductivity sensor and the measured conductivity was proportional to the concentration of the anion in the sample. Recovery rates were between 97% and 104%.

Total phosphorus was determined in two steps. First, an acid digestion with sulfuric acid (CAS no 7664-93-9), which transforms phosphorus into orthophosphate. Secondly, orthophosphate ions were assayed by an automated system. The orthophosphate ion reacted with molybdate (CAS no 12054-85-2) and antimony (CAS no 28300-74-5) ions to form a phosphomolybdate complex. The latter was reduced with ascorbic acid (CAS no 50-81-7) to trigger the appearance of molybdenum blue, whose absorbance at 660 nm is proportional to the concentration of the orthophosphate ion. The detection limit was 200 mg kg⁻¹ and recovery rates were within ±15%.

Total organic carbon was measured by combustion according to method MA. 310-CS 1.0 (Centre d’expertise en analyse environnementale du Québec (CEAEQ), 2013) at 1,360 °C for a maximum of 600 s (oxygen at 30 lb in²). Recovery rates were between 112% and 113%. The water used for the preparation of reagents and standard solutions was distilled or demineralized water. All samples were kept at 4 °C until analysis (less than 14 days).

Community-level physiological profiling

The 96-well microplates contained 31 individual carbon sources from six classes (amine, amino acids, carbohydrates, carboxylic acids, phenolic compounds and polymers) (replicated three times in each plate) along with a clear tetrazolium dye, which gets irreversibly reduced to purple formazan dye when the bacteria consume the carbon (Bochner, 1989). One plate was used for each treatment cylinder (n = 9 carbon sources per treatment). A soil-aqueous solution was made according to methods previously described (Choi & Dobbs, 1999). All optical densities were brought below 0.350 nm (0.264 ± 0.009 nm) through dilutions to normalize the optical density at time zero. The well color development was monitored every day for 6 days by spectrophotometer (590 nm) and raw data was normalized by subtracting the control wells (water) present within each plate’s three replicates. The result were interpreted by the rate of color change in the wells, the area under the curve and the functional richness as described in Choi & Dobbs (1999) and Garland (1997). The average well color development (AWCD) was calculated with the following formula: $\text{AWCD}={\displaystyle \sum {}^{{}^{\left(C-R\right)}}}n$, where C is color production, R is the absorbance of the control (water) and n is the number of substrates (n = 31). To independently estimate color development, a curve-integration (CI) approach was used. The area under the curve (the trapezoid area) was calculated as: ${\displaystyle {\sum }_{i=1}^n}\frac{v_i+v_{i-1}}{2}\times \left(t_i-t_{i-1}\right)$ where v is optical density at time t. Functional richness is defined by the number of positive wells (OD_final−OD_initial > 0.25 nm).

Statistical analysis

Decreases in PHC concentrations, moisture levels, EcoPlates™ data, and temperature between treatments were assessed with a linear mixed-effects repeated measures model fitted by restricted maximum likelihood, performed with the nlme{} (Pinheiro et al., 2017) in R Development Core Team (2016). Post hoc multiple comparison analysis were performed with the Tukey HSD method (lsmeans{} (Lenth, 2016) and mltcompview{} (Hothorn, Bretz & Westfall, 2008)). Assumptions of normality and homoscedasticity were met for all mixed model tests. Gas readings (O₂, CO₂, VOC) were analyzed with the Kruskal–Wallis test by ranks for non-parametric data agricolae{} (De Mendiburu, 2017). Multiple comparisons were corrected with Bonferroni for all parametric and non-parametric tests. Figures were built with ggplot2 (R Development Core Team, 2016; Wickham, 2009).

Soil properties

In the first month, BSG-F temperatures in two bins spiked at least 14 °C above the other bins. Overall, the maximum temperatures reached were 45, 31, 29 and 28 °C for BSG-F, MANR-F, RCW-F and fertilizer treatment (FERT) respectively. There was variability between bins of the same treatment and between treatments, but for the first 90 days, the mean temperature across bins was 22.9 ± 2.7 °C (n = 25,920). Subsequently, the bins’ internal temperatures gradually dropped with the advent of the fall season (Fig. 1A). At this point, BSG-F remained significantly higher than the other treatments (with the exception of the 10-day increment reported at 140 days) (mixed model, p < 0.05). The addition of engine oil (Total Base Number: 10) raised the initial soil pH by around one unit, to approximately 7.4 across all treatments. Subsequently, all treatments remained stable with a pH between 7.2 and 7.6, except for the BSG, (6.8 ± 0.4, after oil addition, and increased to 7.6 ± 0.2 by the end of the experiment). Water requirements varied between treatments. To maintain soil moisture near 70% of field capacity, three watering events took place during the course of the experiment. FERT had little water-holding capacity and needed watering each time. RCW-F required watering during the first two events, but not the third. MANR-F remained within a desirable moisture range (70% of field capacity) for the duration of the experiment and did not require any additional watering. Finally, BSG-F did not require any watering either and it contained the highest moisture levels of all treatments. There was a strong correlation between the amount of moisture (%) and the organic carbon content in the soil (excluding PHC-based carbon). (Fig. 1B).

Nutrient balance

Raw BSG was the amendment richest in nitrogen whereas RCW contained the least nutrients (Table 1). All treatments experienced a sharp decline in available nitrogen ( ${\text{NO}}_2^{-}$, ${\text{NO}}_3^{-}$ and ${\text{NH}}_4^{+}$), which was likely assimilated by soil microorganisms, with the exception of BSG-F, where a sharp increase in NH₃ and ${\text{NH}}_4^{+}$ was observed between June and November. Treatments all show relatively low total nitrogen loss (<15%), except BSG-F that lost nearly 25%. For FERT, an increase in total nitrogen was measured. Since there were no nitrogen inputs over the course of the project, this is probably linked to a heterogeneous distribution of the fertilizer in the soil.

Engine oil decrease

A sharp decrease in the first 82 days, followed by a plateau was observed for all treatments, except the BSG, which maintained a slower but constant reduction over the 5-month experiment (Fig. 2A). There is little variability within each treatment for the initial PHC data before RFMs addition (±57 to ±173 mg kg⁻¹, which represents a variability range of 0.8–2.7%), indicating a homogeneous initial distribution of the engine oil in the soil. However, PHC concentrations were different for all treatments at the start (p < 0.05), which is likely due to uneven dilution from the different RFMs since they were added on a volume basis and PHC concentrations are expressed on a mass basis (Table 1). MANR-F was significantly more effective at driving a PHC reduction (77% ± 3% reduction) in the soil than all other treatments (BSG-F: 38% ± 11%, RCW-F: 60% ± 6% and FERT: 62% ± 4%) (mixed model, p < 0.05). On August 31, while the platform’s usual FERT was fluctuating around the Commercial & Industrial sites’ regulatory guideline (3,500 mg kg⁻¹), MANR-F was well below it and approaching the Residential & Recreational guideline (700 mg kg⁻¹) (Fig. 2A). The analysis of engine oil by fraction revealed the lightest (C₁₀–C₁₆) and heaviest (C₄₀–C₅₀) fractions were either close to or below the detection limit (100 mg kg⁻¹) and were excluded from further analysis. The next lightest fraction (C₁₆–C₂₂) decreased significantly more (63% ± 22%) than the rest of the fractions (C₂₂–C₂₈, C₂₈–C₃₄, C₃₄–C₄₀), which were removed at equivalent rates (42% ± 17%, 39% ± 16% and 44% ± 15%, respectively), despite differences in molecular size (Fig. 2B).

Outgassing

There were notable differences in the gas emissions monitored among the different treatments, despite high variability within the data (Fig. 3). With a mean of 7.9% ± 3.6%, BSG-F contained significantly less O₂ than the other treatments, which all showed averages between 18% and 19% (p < 0.05). CO₂ emissions by FERT and BSG-F were always near the upper detection limit of the equipment (6,000 ppm), while MANR-F was the treatment with the lowest mean emission (2,695 ± 2,871 ppm) (p < 0.05). VOCs were very low in all treatments (means below 250 ppm), except for BSG-F (3,636 ± 3,318 ppm), which released significantly more than all other treatments (p < 0.05). The VOCs emitted in this case could likely be methane from the breakdown of organics in waterlogged pockets of the cylinder which had become anoxic.

Community-level physiological profiling

The global metabolic activity of the plates (AWCD) showed that the soils with amendments were more active than the FERT by the third day and by the fourth day, BSG-F had more AWCD than all others (Fig. 4A) (mixed model, p < 0.05). All treatments followed a sigmoidal shape, presenting a standard bacterial growth curve. The CI approach also revealed that FERT-F was less active than the soils amended with local RFM. There were differences in carbon usage between treatments. For example, BSG-F was able to use polymer carbon sources more effectively than the other treatments and FERT-F used carbohydrates less effectively than all other treatments (mixed model, p < 0.05). Multiple significant differences were also noted within the usage of the carbohydrates, where bacteria in RCW-F were able to actively use more C sources, more rapidly than all other treatments. There was less functional richness in FERT (240 active wells out of 288) than in MANR-F (270), RCW-F (270) and BSG-F (273). Finally, the PCA indicates a clear separation between BSG-F and the other three treatments, which form over-lapping but clear clusters (Fig. 4B). This confirms the different soils’ bacterial community’s catabolic activity fingerprint.

CLPP and soil properties

Community-level physiological profiling conducted with the Ecoplates™ indicated that the biological amendments increased the metabolic activity and functional richness of the soils compared to the fertilizer control (FERT). No well-defined links between PHC-degradation efficiency and Ecoplates-derived metabolic activity were observed. Nonetheless, there were clear clusters in the PCA defining the catabolic capabilities of the soils based on amendment. Microbial activity is optimal between 20 and 40 °C for petroleum remediation and slows when soil temperatures drop close to 0 °C (Battelle & NFESC, 1996). In this project, overall higher temperatures were not linked to higher rates of contaminant removal. Other authors have found that temperature had a great influence on the reduction of petroleum concentrations (Van Gestel et al., 2003), or that higher temperatures could lead to increased volatilization (but unlikely with heavy PHC) (Namkoong et al., 2002). In our case, BSG-F exhibited the highest temperatures, but was the least effective at PHC-removal. This was most likely due to the heat generated during fermentation (anaerobic breakdown) of the BSG. Soil pH is influenced by microbes, and can also be influenced by their activity (Nwankwegu & Onwosi, 2017). Over the course of this experiment, pH remained stable in all treatments, except in the BSG-F. Other studies have reported pH variations in most treatments; this difference in our work could be linked to the soil’s pH buffering capacity or the large-scale set-up, which has different dynamics than microcosms (Nwankwegu & Onwosi, 2017; Prince, 2015). Moisture is an important factor for bioremediation: concentrations too low inhibit microbial activity, and excessive moisture levels can promote hypoxic conditions that are less conducive to rapid biodegradation of PHC (Khan, Husain & Hejazi, 2004). A moisture content of 50–70% of field capacity is often cited as optimal (Agency for Toxic Substances and Disease Registry (ATSDR), 1999; Khan, Zytner & Feng, 2015). In the sandy soil used in the trials, 70% of the field capacity corresponds to about 15% moisture on a mass basis (g g⁻¹). Field capacity varies with soil type and organic matter in soil can improve water retention (Brady & Weil, 1996a). The various local organic amendments used in this project clearly demonstrated this point, with the strong correlation between non-PHC carbon in the soil and moisture levels, and with the different watering requirements of the treatments. FERT had virtually no organic matter and required moisture addition throughout the project. RCW-F required watering the first two events, but not the third, indicating that the wood had started to retain water in the soil. BSG-F retained moisture slightly above levels for optimal degradation (20% ± 1.7% (g g⁻¹) at the start), but the aspect of most concern was that this high moisture was coupled with a fine particle and compact amendment structure which limited air circulation in the bins, fostering hypoxic conditions which are not ideal for rapid PHC degradation. Further, the BSG supplied soil microbes with an abundance of easily accessible food which may have been easier to target than the engine oil. Finally, MANR was the optimal RFM input for moisture since MANR-F did not require any watering throughout the project, while maintaining values close to 70% field capacity. Reduction in soil management operations can be advantageous in large-scale settings.

Nitrogen balance and engine oil removal

In this study, nitrogen came from two main sources: an inorganic fertilizer and the three RFMs (soil nitrogen content was negligible). No nitrogen was added during the course of the study. We had aimed to keep the soils under aerobic conditions where volatile forms of N are not normally emitted, hence total nitrogen should have remained stable throughout the monitoring. It did, with the exception of BSG-F where a nearly 25% drop in nitrogen occurred. Other than heterogeneous distribution in soil which can skew the results, total nitrogen loss in soils may be related to leaching of soluble forms (although bins were largely protected from rainfall) or a denitrification process resulting in the formation of nitrogen gas and its loss to the atmosphere. This last process may indicate conditions in soils that are too low in oxygen. The reduced forms of nitrogen ( ${\text{NH}}_4^{+}$ and ${\text{NH}}_3$) present in the BSG-F soil in November support the presence of sub-optimal oxygen levels since these reduced forms of nitrogen are usually observed when the soil is deficient in oxygen. Whereas, when the soil is well aerated, nitrogen is found in oxidized forms such as nitrate and nitrite (Brady & Weil, 1996b). There was an initial input of ammonium as diammonium phosphate in the inorganic fertilizer, but this was not detected in the BSG bins at the beginning of the experiment. The concentrations detected in November were likely mineralized from the organic matter in the amendment and remained in this form due to lack of oxygen for conversion to oxidized forms (NO₃⁻ and NO₂⁻) by the nitrification process. Reduced forms of nitrogen are most prevalent under hypoxic soil conditions (Britto & Kronzucker, 2002) and effective aliphatic PHC-degradation usually occurs under aerobic conditions (Khan, Husain & Hejazi, 2004).

Larger PHC molecules were expected to be biodegraded at slower rates than the lighter ones, but this was not generally the case in our study (Battelle & NFESC, 1996; Khan, Husain & Hejazi, 2004). The C₁₆–C₂₂ fraction was indeed removed faster than the other fractions monitored, but the decrease of the other three fractions (between C₂₂ and C₄₀) was statistically equivalent (p > 0.05). PHC can move out of soil through volatilization, but other authors found mainly fractions below 12–16 carbons to be volatile while heavier compounds were mainly biodegraded (Namkoong et al., 2002; Gallego et al., 2010). Bacteria usually reduce PHC molecules down to CO₂ (Brady & Weil, 1996c). After 82 days of the different treatments, MANR-R had significantly decreased PHC concentrations and was approaching the Residential and Recreational guideline (700 mg kg⁻¹), while the LTU’s usual FERT treatment was fluctuating around the Commercial and Industrial guideline (3,500 mg kg⁻¹) of the Quebec government regulations. From this point on to the end of the experiments all treatments plateaued, and little degradation took place in the last 65 days. Since varying levels of carbon from the PHC remained in the soil, it would be possible to optimize the treatments to stimulate degradation beyond the plateaus. This could be done through better control over parameters such as aeration, moisture and nutrients, and using known correlations for estimating petroleum biodegradation rates in soils, which are especially useful for scaling-up experiments (Khan, Zytner & Feng, 2015).

BSG-F did not exhibit the same plateau but was far less effective than the other treatments for engine oil removal (38% ± 11%). Other authors have found BSG to be an effective soil amendment. In a phytoremediation experiment using the plant Jatropha curcas, Agamuthu, Abioye & Aziz (2010) found that when BSG was added to a soil contaminated with 2.5% and 1% of lubricating oil, the degradation rates of the PHCs were 89.6–96.6% respectively. Oruru (2014) demonstrated not only the efficiency but also the sustainability of BSG as an amendment to treat diesel-contaminated soil. The reduced efficiency of the BSG-F in this study appears to be linked to the soil structure and excess moisture which restricted aeration. Despite its lower PHC-removal performance in this project, the application of BSG could be further considered for the remediation of soils with low nutrients and low water retention capacities, as long as aerobic conditions are maintained to minimize nitrogen loss and VOC output. Currently, farmers supplement their animals’ regular feed with BSG, but rapid disposable is primordial for breweries because it ferments rapidly causing serious odor problems. This leads many city-based breweries to send their BSG to the landfill (Santos et al., 2003).

RCW-F was equivalent to FERT for PHC-removal and its measured contributions within this project are limited to improvements of soil parameters (moisture and microbial activity). MANR-F was unequivocally the most successful soil amendment in this project. It led to the most PHC-reduction in the shortest time period. From a soil clean-up facility’s perspective, this represents a faster turnaround of the soil and potential increases in profitability. MANR-F maintained adequate moisture and oxygen levels, while releasing the least amount of CO₂, which is interesting in a global climate change context. The bacterial degradation of PHC emits CO₂ and since the most PHC-removal took place in this treatment, it should have released more than the rest. Since it was not the case, we hypothesized a form of carbon sequestration in the microbial biomass, but this was not tested further.

Cost of RFM and opportunities for circular economy?

Within the scope of this project, MANR was available nearby and free of cost, but in other regions such manure may be sold at a higher price or used directly on adjacent farmlands, potentially reducing the application of our results to some regions. RCW did not significantly increase remediation speed compared to fertilizer alone, but it promoted long term moisture retention, which can alleviate the management of biopiles through reduced watering needs. BSG led to less PHC removal than the LTU’s usual fertilizer treatment by limiting aeration due to excess moisture. In retrospect, the addition of 30% BSG for this soil was excessive. We hypothesize that a smaller proportion of the dense and nutritious BSG, coupled with the addition of a bulking agent such as RCW could create favorable PHC remediation conditions at an advantageous cost. We conclude that there is no one-size-fits-all approach for the use of RFMs, and each remediation site will need to evaluate the RFMs available in their respective region.

In their Vision 2050 project, the World Business Council for Sustainable Development envisions making the concept of waste obsolete as a normal business practice (World Business Council for Sustainable Development (WBCSD), 2010). This research highlights a particularly interesting opportunity for soil treatment industries to actively participate in this vision, by up-cycling degraded “waste soil” through remediation and introducing it into a circular economy loop. Further, RFM are present across the globe and can offer low-cost amendments to boost remediation efficiency while reducing treatment time.