High-temperature electrothermal remediation of multi-pollutants in soil

Soil contamination is an environmental issue due to increasing anthropogenic activities. Existing processes for soil remediation suffer from long treatment time and lack generality because of different sources, occurrences, and properties of pollutants. Here, we report a high-temperature electrothermal process for rapid, water-free remediation of multiple pollutants in soil. The temperature of contaminated soil with carbon additives ramps up to 1000 to 3000 °C as needed within seconds via pulsed direct current input, enabling the vaporization of heavy metals like Cd, Hg, Pb, Co, Ni, and Cu, and graphitization of persistent organic pollutants like polycyclic aromatic hydrocarbons. The rapid treatment retains soil mineral constituents while increases infiltration rate and exchangeable nutrient supply, leading to soil fertilization and improved germination rates. We propose strategies for upscaling and field applications. Techno-economic analysis indicates the process holds the potential for being more energy-efficient and cost-effective compared to soil washing or thermal desorption.


Supplementary Note 1. Influence of chemical species on removal efficiencies of heavy metals
In the HET process, the elevated temperatures can initiate various reactions, including carbothermic reduction facilitated by the presence of carbon conductive additives, thermal decomposition, and evaporation. Consequently, there are multiple potential pathways for the removal of different heavy metals species: (1) Direct evaporation of metal species; (2) Thermal decomposition of metal species to other species followed by their evaporation; and (3) Carbothermic reduction of metal species to lower-valence-state species followed by their evaporation. Given that the temperature in the HET process can reach up to 3000 °C, all of the above reactions have the potential to occur, providing various pathways for the removal of different heavy metal species.
In this study, we focused on Hg as an example to investigate the impact of speciation on removal efficiency. The Hg contaminants present in the soil primarily exist in the form of Hg (0) and Hg(II) (ref 1 ). Depending on the counterions, Hg(II) species can include HgS, HgO, HgCl2, HgSO4, and so on 1 . To understand the potential reactions involved, we conducted a thermodynamic analysis using the software HSC Chemistry 10.
Next, we used the HET process to remove Hg, HgO, and HgSO4 from contaminated soil.
These individual Hg species were added separately to the soil, which was then mixed with carbon black as conductive additives. The HET conditions remained consistent (100 V, 1 s) for all Hg contaminants ( Supplementary Fig. 16c). Subsequently, we measured the removal efficiencies of each Hg species. As shown in Supplementary Fig. 16d, the HET process achieved high removal efficiencies for all Hg species: Hg (~90.4%), HgCl2 (~94.6%), HgO (~95.1%), and HgSO4 (~86.5%) using a single HET pulse of 1 s. It is worth noting that HgCl2 possesses a higher vapor pressure compared to Hg ( Supplementary Fig. 16b), leading to its slightly higher removal efficiency. Additionally, HgO is prone to decomposition into Hg ( Supplementary Fig. 16a), thereby exhibiting a high removal efficiency. On the other hand, the decomposition of HgSO4 is relatively more challenging and occurs at higher temperatures ( Supplementary Fig. 16a), resulting in a slightly lower removal efficiency compared to the other Hg species.

Supplementary Note 2. Strategy for scaling up the HET process.
To demonstrate the scalability of the HET process, we first conducted a theoretical analysis of the scaling rule of the HET process. Then, we performed batch-by-batch scaling up experiments in our research lab, with the productivity up to kg scale per day. Next, we proposed an ex-situ prototype for continuous HET processing using a belt roller. Finally, we provide a conceptual design of a tractor attached HET unit as well as a field test facility for in-site soil remediation.

Scaling rule of HET process by theoretical analysis.
Achieving effective removal of heavy metals and organic contaminants through the HET process largely depends on the maximum temperature reached. Therefore, ensuring consistent temperature across the sample is critical when scaling up the process. In Joule heating, the amount of heat (Q) can be calculated by Supplementary Equation 9, where I is the current passing through the sample, R is the resistance of the sample, and t is the heating time. The amount of heat per unit volume (Qv) can then be determined by Supplementary where j is the current density, ρe is the electrical resistivity of the sample, and t is the heating time.
where ρm is the sample density, S is the sample cross-sectional area, and L is the sample length. ρm is constant considering the same compression of the sample.
We can further obtain Supplementary Equation 17 to determine the current density, = m (17) As discussed earlier, maintaining a constant current density (j) is necessary to increase the sample mass (m), which can be achieved through practices including: (1) linearly increasing the HET voltage (V), and (2) linearly increasing the capacitance (C).

Demonstration of the scaling of the HET process in our research lab.
In our first-generation HET system, the capacitor bank is made up of 10 commercial aluminum electrolytic capacitors (450 V, 6 mF, Mouser #80-PEH200YX460BQU2) with a total capacitance of C0 = 0.06 F. In our small-scale experiment (Supplementary Table 2), we used a sample mass of m0 = 0.2 g and HET conditions of V0 = 100 V and C0 = 0.06 F, resulting in a temperature of ~3000 °C (Fig. 1d). Here, we have successfully scaled up the HET to a larger sample mass of m1 = 2 g ( Supplementary Fig. 29). For this purpose, we built a second-generation HET system For the sample mass of m1 = 2 g and C1 = 0.624 F ( Supplementary Fig. 29b), we used a HET voltage of V1 = 120 V, which fit the Supplementary Equation 18. Since the achievable temperature is critical for the heavy metal removal by evaporation and organic contaminants removal by graphitization, the temperature for the large-scale sample was recorded ( Supplementary Fig. 29c).
The maximum temperature also reached ~3000 °C, demonstrating the successful scaling up of the HET process. The heavy metal concentration in the r-Soil was reduced, and the removal efficiency of heavy metals was calculated to be 40 to 80% for one-time HET ( Supplementary Fig. 29d), comparable to that of the small-scale sample (Fig. 2c).
In addition, we have further upscaled the sample mass to ~8 g per batch using Metcoke as the conductive additive ( Supplementary Fig. 29e), and achieved a total treated soil mass of ~100 g with processing time of <10 min after sieving separation of the Metcoke (Supplementary Fig. 29f).
As a result, we are currently able to achieve a production rate of >10 kg day -1 in our research lab.

Use of the AC-HET process for scaling up to 100 g per batch.
We integrated an alternating current (AC) supply with the HET process, which offers better scalability compared to the direct current (DC) process ( Supplementary Fig. 30). We have extended the AC supply to treat a sample mass of 100 g per batch ( Supplementary Fig. 31). Pyrenecontaminated soil was used as a representative. The soil was mixed with Metcoke as the conductive additive and loaded into a quartz tube with an inner diameter of 4 cm ( Supplementary Fig. 31a).
The sample was connected to the AC system for thermal treatment ( Supplementary Fig. 31b). After treatment, the mixture of treated soil and Metcoke was obtained ( Supplementary Fig. 31c), and the Metcoke was removed by sieving, resulting in separated treated soil ( Supplementary Fig. 31d).
After two cycles of AC-HET treatment, the pyrene concentration in the soil was reduced to below the safe content ( Supplementary Fig. 31e-f), demonstrating the effectiveness of the HET process for the removal of polycyclic aromatic hydrocarbons (PAHs). The treatment process took approximately 1 minute for the 100 g soil treatment. This production rate corresponds to ~6 kg h -1 or ~144 kg day -1 .
The HET process can be integrated with various industrial scale-up technologies. Here, we introduce a prototype assembly of the HET process with a belt roller for continuous processing ( Supplementary Fig. 32a). In this assembly, the c-Soil/CB mixture is loaded into a chamber, compressed to proper resistance, undergoes the HET process, and finally, the remediated soil is unloaded. We note that this is just one possible method for continuous processing, and other established industrial scaling techniques could also be applied.
In another design ( Supplementary Fig. 32b), the HET process can be conducted in a continuous manner. The thin and flexible sheet metal belt is used for soil conversion. As the sheet metal is not a good electrical conductor along its length, the current is concentrated where the electrodes are located, i.e., the flash zone. The top electrode is segmented longitudinally, consisting of a stack of plates with thin insulation separating them. If there is non-uniform resistance in the sample, the current may become localized, resulting in parts of the sample intensely heated, and other higher resistance parts minimally heated. In this scheme, the current in each wire to each segment will be the same, as each will have a separate current control, such as an insulated-gate bipolar transistor (IGBT) or zero-crossing relay if AC power is used. A resistor divider circuit is a low-cost option but less energy-efficient. Temperature sensors can be embedded in the tip of each electrode segment to assure that all parts of the moving sample are heated equally. The wire leads are shown separately for visualization, but would all be in a row perpendicular to the plane of the drawing plane. This allows for an indefinite width of the belts by adding more electrode segments.

Conceptual design of a tractor attached HET unit for on-site soil remediation.
Considering the high cost of excavating and transporting contaminated soil to a remediation facility, as well as the difficulty in providing electricity to remote areas, we here propose a conceptual design for on-site soil remediation using a tractor-attached HET unit ( Supplementary Fig. 33). In this design, the HET is powered by a diesel generator or rechargeable batteries. The process involves the following steps: (1) Disc ploughing the soil to soften it. (2) Excavating and converting the contaminated soil using a sheet metal belt and tensioning roller system. (3) Drying the soil to reduce moisture content, if necessary. (4) Adding carbon additives and mixing them with the dried soil. (5) Compressing the mixture to an appropriate resistance. (6) Joule heating the mixture using electricity provided by the generator and collecting the heavy metal volatiles in a trap. (7) Separating the carbon additives from the soil by sieving. (8) Redepositing the remediated soil at nearly its original location.
In addition, the HET process can be integrated with traditional thermal desorption method for soil remediation. We also propose a modified facility design based on a known method 5 , but in our case the electrodes provide a rapid voltage pulse for electric heating rather than longduration heat injection ( Supplementary Fig. 34). The facility includes a vacuum well, vacuum piping, collector, filtration, blower, and exhaust stack. In this design, the contaminated soil is considered as a dry, porous material. Soil is firstly mixed with carbon conductive additives. The electrodes are installed in the heating well, with the depth determined by the level of contamination.
A vacuum collection system is designed to capture the volatile matter during the HET process.

Consideration of the soil moisture.
We have also tested the applicability of the HET process for remediation of moisture-containing soil. We measured the moisture content of the soil used in our experiments. The previously used dry soil had a moisture content of ~1.2% ( Supplementary Fig. 35a). We collected another batch of soil with a moisture content of ~14% (denoted as Moisture soil, Supplementary Fig. 35b). Pyrene, a representative contaminant, was added to the moisture soil, which was then mixed with carbon black as the conductive additive. The electric input of the HET treatment was the same as that for the dry soil (Supplementary Table 2 The pyrene removal efficiency in the moisture soil reaches 91% after 3 HET pulses, which is slightly lower than that of the dry soil (95%). This demonstrates the feasibility of the HET process for remediation of moisture-containing soil.

Supplementary Note 3. The electrical energy cost evaluation of the HET process.
The energy consumption is calculated using Supplementary Equation 19, Where E is the energy per gram (kJ g -1 ), V1 and V2 are the capacitor voltages before and after HET, respectively, C is the capacitance (60 mF), and M is the mass per batch. This value is consistent with the small-scale sample. Theoretically, the electric energy density plays a crucial role in the HET remediation process. As long as the energy density remains the same during the scaling-up process, the estimation of energy consumption will remain consistent.
We conducted a cost comparison between the HET purification process and established thermal remediation methods, which use heat to eliminate contaminants through thermal desorption, destruction, or immobilization. Traditional thermal remediation techniques include thermal conduction heating (TCH), steam-enhanced extraction (SEE), electrical resistance heating (ERH), and radio frequency heating (RFH) 6 . The energy consumption of the HET process is comparable to or lower than these traditional methods (Supplementary Table 3, Supplementary   Fig. 36), despite the fact that the HET process allows for a much higher temperature and thus can remediate a wider range of contaminants with less volatility. Furthermore, even though the HET process operates at a higher temperature than traditional thermal methods, its processing time is significantly shorter, resulting in similar or lower energy consumption. In addition, the HET process is more energy-efficient than other innovative electricity-based remediation techniques such as the electrochemical process 7 .

Goal and scope
This study is conducted in accordance with the requirements outlined in ISO 14044 (ref 8 ).The primary objective is to assess the environmental and energy impact, as well as the cost, of the HET process for soil remediation. Specifically, the analysis aims to determine whether the newly established HET process reduces energy consumption and water consumption compared to other soil remediation methods 9 such as thermal desorption 10 , soil washing 11 , and chemical degradation 12 . Additionally, the study evaluates the cost competitiveness of the HET process in comparison to these established soil remediation techniques.

Scenario description and system boundaries
In this study, four scenarios were examined ( Supplementary Fig. 37 Scenario 2 Thermal Desorption. In this scenario, a typical thermal desorption process is employed 13 , where 1 tonne of contaminated soil is treated at 350 ℃ for 90 min using a furnace. Scenario 3 Soil Washing. In this scenario 11 , 1 tonne of contaminated soil is extracted with a 4000 L of mixture solvent (5% pentanol-10% water-85% ethanol) using a rotating shaker at 16 rpm for 1 h. The treated soil/solvent is then separated by filtration. The soil washing process is conducted for 3 cycles, consuming 600 L of 1-pentanol (0.488 tonnes), 1200 L of water (1.2 tonnes), and 10200 L of ethanol (8.048 tonnes). It is assumed that the solvents can be recovered, with a recovery yield of 98%, although the specific recovery process is not accessed in this study.
Scenario 4 Chemical Oxidation. In this scenario 12 , 1 tonne of contaminated soil is mixed with a mixture solution containing 3.22 tonnes of H2O2 (30%) and 85.5 tonnes of H2O. Then, the mixture is agitated at 200 rpm for one week. Finally, the soil and wastewater are separated by filtration. It is assumed that the chemicals used in this process can be recovered, with a recover yield of 90%, although the recovery process is not accessed in this study.

Life-cycle inventory
The water consumption and energy consumption values for raw materials and processing are summarized in Supplementary Table 5, which are explained in detail below.

Raw materials.
The values for coke from coal production for steel manufacturing, H2O2 (30%), and ethanol production from corn-wet milling corn ethanol are obtained from US department of energy (DOE) GREET Model 14 . Ethanol is used as a proxy for 1-pentanol since 1-pentanol is not included in the GREET databases.
Processing -Mixing. Energy input is required for the mixing of conductive carbon with contaminated soil. It is assumed that the mixing is conducted using an electricity-driven Powder Mixer 15 , with an estimated energy consumption of 9.432 MJ tonne -1 . No water consumption is involved in this process.
Processing -HET. Energy consumption for the HET process is estimated to be 1512 MJ tonne -1 (Supplementary Note 3). No water consumption in associated with this process.
Processingsieving. The separation process is assumed to be conducted using an industrial vibrating sieving system. For an electrical shaker machine, the estimated energy consumption is ~4 MJ tonne -1 . No water consumption occurs during the sieving process.
Processing -Furnace heating. In the case of thermal desorption, it is assumed that furnace heating is carried out using an electrical furnace 16 , with a load of 45 kg, power of 10 kWh, and heat-up time of 2 h. The treatment time is 90 min. The energy consumption is estimated to be ~778 kWh tonne -1 , or 2800 MJ tonne -1 . This value is comparable to the reported energy consumption values for thermal desorption found in literatures ( Supplementary Fig. 36, Supplementary Table   3).

Processing -Agitation.
It is assumed that an industrial agitator 17 with a power of 0.4 kW, tank capacity of 0.5 m 3 , and operating time of 1 h, is used. The energy consumption is calculated to be 0.8 kWh tonne -1 , or 2.88 MJ tonne -1 .

Processing -Filtration.
The filtration process is assumed to be conducted using a filtration system. The energy consumption is estimated to be ~2.2 MJ tonne -1 .

Life-cycle impact assessment
In this study, the environmental impacts are assessed using two indicators, cumulative water use (Supplementary Table 6) and cumulative energy demand (Supplementary Table 7).

Cost evaluation
In this study, the cost of raw materials is based on the prices of commercial products,  Fig. 29) is ~$5000, with an Annual Production of ~30 tonnes (assuming 10 g per batch with a treatment duration of 10 s). The Capital Expense is calculated by: Assume a Useful Lifetime ranging from 10 to 20 years, the Capital Expense is calculated to be $8.35 to $16.7 tonne -1 . The Total Expense is then calculated by: The Total Expense of the HET process ranges from $51.7 to $60 tonne -1 .
Comparatively, the typical Total Expense of ex-situ thermal desorption ranges from $46 to Expense is calculated to be $9 to $18 per tonne of contaminated soil. The total Expense of the chemical oxidation process is estimated to be $172 to $181 per tonne of contaminated soil.

Sensitivity and Uncertainty.
There are several important considerations and potential sources of uncertainty in this study. Firstly, the water consumption, energy consumption, and price data for raw materials used in this study were obtained from different sources, which may introduce some uncertainty.

CAUTION:
There is a risk of electrical shock if improperly operated. We recommended the below safety guidance when using this equipment. More safety practices could be found in our previous publications 27,28 .
1. Enclose or carefully insulate all wire connections.
2. All connections, wires, and components must be suitable for the high voltages and currents.
3. One hand rule. Use only one hand when working on the system, with the other hand not touching any grounded surface.
4. Provide a mechanical discharge circuit breaker switch connected to a power resistor of a few hundred ohms to rapidly bleed off the capacitor charge.
5. Provide a "kill" circuit breaker switches to disconnect the sample holder from the capacitor bank.
6. Post high voltage warning signs on the apparatus. 7. Keep in mind that the system can discharge many thousands of Joules in milliseconds, which can cause components such as relays to explore.

21
The concentrations of heavy metals in clean soil are low (Cd undetectable, Hg undetectable, Pb ~0.6 ppm, Co ~4.5 ppm, Ni ~30 ppm, and Cu ~79 ppm), therefore the concentration of heavy metals in the contaminated soil are controlled by spiking with metal salts (Cd ~100 ppm, Hg ~300 ppm, Pb ~1000 ppm, Co ~2000 ppm, Ni ~10000 ppm, and Cu ~10000 ppm). The concentrations of heavy metals in carbon black (Cd ~17 ppm, Hg undetectable, Pb ~10 ppm, Co undetectable, Ni ~6 ppm, and Cu undetectable) are far below that of the contaminated soil (Cd ~100 ppm, Hg ~300 ppm, Pb ~1000 ppm, Co ~2000 ppm, Ni ~10000 ppm, and Cu ~10000 ppm), and hence will not introduce significant error during the HET process. ppm, Co ~200 ppm, Ni ~10000 ppm, and Cu ~10000 ppm), and hence the use of Metcoke as conductive additive during the HET process will not introduce significant error. The removal efficiencies of heavy metals with Metcoke as an additive are ~60% for most of the metals, slightly lower than those with carbon black as an additive (Fig. 2c). The difference may be due to the higher conductivity of carbon black compared to Metcoke (R ~1.0 Ω for CB as an additive, and R ~2.0 Ω for Metcoke as an additive), resulting in better removal efficiencies. Additionally, the smaller particle size and higher surface area of carbon black could result in more homogeneous heating, providing better removal efficiencies than Metcoke.
Flash graphene (FG) was also used as a conductive additive . The FG is synthesized using Metcoke as the precursor. The concentrations of heavy metals in the FG are as below: Cd undetectable, Hg undetectable, Pb undetectable, Co ~1.1 ppm, Ni ~8.6 ppm, and Cu ~47 ppm.
These values are slightly lower than those in the raw materials due to evaporative loss of heavy metals during the FG synthesis. In addition, these values are well below the contaminated soil levels (Cd ~100 ppm, Hg ~200 ppm, Pb ~1000 ppm, Co ~200 ppm, Ni ~10000 ppm, and Cu ~10000 ppm), posing no significant error during the HET process. The removal efficiencies of heavy metals are >60%, which is slightly less than that of carbon black as an additive (Fig. 2c).
The reason might be similar with that of the Metcoke as an additive: firstly, carbon black has superior conductivity to FG (R ~1.0 Ω for CB as an additive, and R ~1.5 Ω for FG as an additive); Secondly, carbon black has a much smaller particle size and much higher surface area than FG.
Furthermore, plastic pyrolysis ash (Plastic Ash) was tested as a conductive additive.  efficiencies of heavy metals are >40%, which is slightly lower than that achieved by using carbon black as an additive (Fig. 2c). This may be due to similar reasons as with the Metcoke as a conductive additive: firstly, carbon black has better conductivity than Plastic Ash (R ~1.0 Ω for carbon black as an additive, and R ~3.0 Ω for Plastic Ash as an additive); secondly, carbon black has a smaller particle size and a higher surface area than Plastic Ash. Supplementary Fig. 9. Carbon residue in the r-Soil. TGA curve of the r-Soil with residual carbon. TGA was conducted in air with the heating rate of 10 °C min -1 .
During the HET process, the easy-to-decompose components in the soil were decomposed. Hence, in the TGA measurement, most of the weight loss was ascribed to the residual carbon. Based on the density and particle size difference between soil and the conductive additives, we were able to separate them by physical processes. By using metallurgical coke (Metcoke) as an example, we demonstrated the separation and reuse of the conductive additive. We selected Metcoke with a particle size larger than that of soil ( Supplementary Fig. 10a). After the HET purification process, the particle size of the Metcoke remained larger than that of the soil ( Supplementary Fig. 10b). Hence, we were able to separate the soil and Metcoke by sieving The Raman spectra showed that the Metcoke was transferred to flash graphene after the HET process. A mixture of soil (~200 mg) and BAC (~100 mg) were used for the HET process. In a typical experiment, the recovered mass of BAC was m(recovered BAC) = 95.5 mg, resulting in a BAC recovery yield of ~95.5%. The recovered BAC could be reused for further HET treatment.
To demonstrate this, we used the recovered BAC (95.5 mg) with some new BAC (4.5 mg) as the conductive additives to purify another batch of soil (200 mg). After the HET process and subsequent separation by sieving, we recovered the BAC with a mass of m(recovered BAC) = 93.2 mg, resulting in a BAC recovery yield of ~93.2%. Supplementary Fig. 14. XPS characterization of heavy metals. (a, b) XPS fine spectrum of Cu The contaminated soil was mixed with biochar at a mass ratio of 2:1. The HET treatment conditions for this mixture remained the same as those used for carbon black additives (Supplementary Table 2). Following the HET treatment, the pyrene content was significantly reduced ( Supplementary Fig. 20c). The removal efficiency of pyrene achieved by a single HET pulse reached ~95% ( Supplementary Fig. 20d), which is comparable to the performance observed with other carbon additives (Fig. 3). To determine whether the PAH are removed from the soil by graphitization or evaporation, we constructed an apparatus to collect the evaporative PAH during the HET process. The setup, shown in Supplementary Fig. 22a, consisted of a cold trap connected to the HET chamber via a vacuum tubing with a porous Cu lead to permit gas diffusion. The cold trap was first pumped to vacuum. Then, the HET was conducted, with the unreacted PAH remaining in the r-Soil, or evaporated and condensed on the quartz tube, the Cu lead, the vacuum tubing, or in the cold trap.
The vacuum was released, and the PAH weights in the r-Soil, quartz tube and Cu lead, vacuum tubing, and cold trap were measured by the same solvent extraction process. The unaccounted PAH weight was considered as the graphitized fraction. As shown in Supplementary Figs. 22b-d, most of the PAH were graphitized, demonstrating that the organic contaminants in the contaminated soil were predominantly removed through graphitization rather than evaporation.
Supplementary Fig. 23. Particle size distribution measurement. The particle size distribution of raw soil and HET treated soil measured using laser particle size analyzer.  Note: a The environmental impacts or energy demands are normalized to production or processing of 1 tonne of materials. b Ethanol was used as a proxy for 1-pentanol because 1-pentanol was not included in the GREET database.   Note: a The materials mass flow is normalized to remediation of 1 tonne of PAH-contaminated soil. Note: a The materials mass flow is normalized to remediation of 1 tonne of PAH-contaminated soil.