Urban mining by flash Joule heating

Precious metal recovery from electronic waste, termed urban mining, is important for a circular economy. Present methods for urban mining, mainly smelting and leaching, suffer from lengthy purification processes and negative environmental impacts. Here, a solvent-free and sustainable process by flash Joule heating is disclosed to recover precious metals and remove hazardous heavy metals in electronic waste within one second. The sample temperature ramps to ~3400 K in milliseconds by the ultrafast electrical thermal process. Such a high temperature enables the evaporative separation of precious metals from the supporting matrices, with the recovery yields >80% for Rh, Pd, Ag, and >60% for Au. The heavy metals in electronic waste, some of which are highly toxic including Cr, As, Cd, Hg, and Pb, are also removed, leaving a final waste with minimal metal content, acceptable even for agriculture soil levels. Urban mining by flash Joule heating would be 80× to 500× less energy consumptive than using traditional smelting furnaces for metal-component recovery and more environmentally friendly.


Supplementary Notes
Supplementary Note 1. The Numerical simulation.
The numerical simulation was conducted using the finite element software COMSOL Multiphysics 5.5.

Gas diffusion simulation.
In a typical FJH process, the vessel was first pumped to P0 ~10 Pa. After a typical FJH, the pressure was P1 = 12 kPa, hence the collected gas pressure was ΔP1 = 12 kPa.
The volume of the vessel is V1 ~40 mL, and the volume of the quartz tube is V2 ~1 mL.
According to the Boyle's Law, eq S1: The inner pressure (ΔP2) was calculated using eq S2: Hence, the inner pressure generated during the FJH heating was set as 5 atm for the gas diffusion simulation. 3 For the gas diffusion simulation, the Laminar Flow mode was used with the following conditions ( Supplementary Fig. 18): (1) Geometrical parameters: reactor radius (0.4 cm), reactor length (2 cm), tube radius (0.1 cm), tube length (4 cm).

Supplementary Note 2. The calculation of recovery yield
For the evaporative separation, considering that the mass of PCB raw materials used for FJH is m(PCB), the concentration of precious metals in PCB raw materials was measured as c (PCB), and the mass of precious metals condensed in the cold trap was M(Gas), the recovery yield (Y(Gas)) was calculated by as in eq S3:

M(Gas) + c(PCB-Flash) × m(PCB-Flash).
In this case, eq S1 was used. However, in some cases, due to the inhomogeneous distribution of precious metals in PCB raw materials, the concentration of precious metals in the PCB raw materials used in different batches had some variation. This could result in a recovery yield >100% for the first method. In this case, eq S2 was used to give a lower limitation of recovery yield.
For the FJH-improved leaching efficiency, considering that the mass of PCB raw materials used for FJH was m(PCB), the concentration of precious metals in PCB raw materials was measured as c(PCB), the mass of PCB-Flash solid was m(PCB-Flash), the concentration of precious metals in PCB-Flash solid was measured as c(PCB-Flash), then the recovery yield by leaching the PCB-Flash solid, Y(PCB-Flash), was calculated using eq S5: Similarly, the recovery yield by leaching the PCB-Calcination, Y(PCB-Calcination), was calculated using eq S6: where m(PCB-Calcination) was the mass of PCB-Calcination solid, the c(PCB-Flash) was the concentration of precious metals in PCB-Calcination solid.
The recovery yield by leaching the PCB-Flash-Calcination, Y(PCB-Flash-Calcination), was calculated using eq S7: where m(PCB-Flash-Calcination) was the mass of PCB-Calcination solid, the c(PCB-Flash-Calcination) was the concentration of precious metals in PCB-Calcination solid.
For the removal of toxic heavy metals by FJH, considering that the mass of PCB raw materials used for FJH was m(PCB), the concentration of heavy metals in PCB raw materials was measured as c(PCB), the mass of PCB-Flash solid was m(PCB-Flash), the concentration of precious metals in PCB-Flash solid was measured as c(PCB-Flash), then the removal efficiency, Y(Removal), of toxic heavy metals by FJH was calculated using eq S8: Furthermore, considering that the mass of toxic heavy metals collected by condensation in the cold trap was M(Gas), the collection yield, Y(Collection), was calculated using eq S9: Supplementary Note 3. The total composition analysis of the collected solid and evidencebased predictions on purification and refining.
The collected solids by the evaporative separation are a mixture of metals. There are already many commercial processes to separate individual metals from a mixture of metals.
Here, we first did a complete composition investigation of metals (type and content) in the collected solids and provide evidence-based predictions on how to separate individual metals using readily accessible, well-established methods.
The total composition analysis of the collected solid.
We use the collected solid without chemical additive, and the collected solid with mixture halide additives (NaF, NaCl, and NaI) as representatives. The abundant metals in e-waste (Cu, Sn, Al, Fe, and Zn), the precious metals (Rh, Pd, Pt, Ag, and Au), and the toxic heavy metals (Hg, Cd, As, Pb, and Cr) were measured. Since the content of other metals are a few orders of magnitude less than those of the abundant metals, we think that the consideration of the above elements affords a reasonable approximation. The collected solid was totally digested and the ICP MS analysis of the metals was conducted. The total metal composition of the collected solids with and without the additives was shown in Supplementary Fig. 14. In both cases, the most abundant components are Cu with >60 wt%, followed by Al, Sn, Fe, and Zn with >1 wt%.
For precious metals, the mass ratios were Ag, ~0.6 wt%; Pd, ~0.04 wt%; Rh, ~0.02 wt%; and Au, ~0.01 wt%. Cu is one major metal to be recovered with ~30% of total values for urban mining, so our FJH process also works for Cu recovery from e-waste.
The chemical states of precious metals in the collected solid.
Since the content of precious metals are <0.1 wt% ( Supplementary Fig. 14), it is difficult to directly conduct the XPS analysis. Here, we added precious metal salts into the e-7 waste and conducted the same FJH procedure to collect the volatiles for the XPS analysis.
Specifically, RhCl3, PdCl2, AgCl, and HAuCl4 were added into the mixture of e-waste and carbon black (CB) with a weight ratio of 5 wt% for each. After the same FJH process, the volatiles were collected and XPS measurements were conducted to analyze the chemical states of the precious metals. As shown in Supplementary Fig. 16 Evidence-based prediction on the purification and refining of precious metals.
There have been a few commercial processes for individual precious metals separation and refining, including selective precipitation, solvent extraction and solid-phase extraction 1,2 .
The classical precipitation methods are based on the solubility difference of the ammonium salts of precious metal-chloro complexes 3 . The solvent extraction methods use solvent extractants to selectively extract a given metal and then transfer to the organic phase 4 . The solid-phase extraction relies on the use of ion-exchange resins with metal-selective ligands 5 .
All these well-established techniques could be directly used in our collected solid for purification and refinement of individual precious metals.

Supplementary Note 4. Separation capability of the evaporative separation process.
Obtaining readily applicable pure metals from the complex e-waste usually relies on a lengthy engineering package, including beneficiation to purification. It usually has two major processes, the recovery of metal mixture from the e-waste raw materials, and the subsequent separation or refining of individual metals from the mixture. In commercial practices, after examination and beneficiation of the e-waste, the pyrometallurgical process (smelting) is applied to obtain a mixture solid of metals. Then the hydrometallurgical process (leaching) is used to obtain the leaching liquor with mixed metals. Finally, advanced refining process are used to separate and purify individual metals, with the main techniques including solvent extraction, leaching-precipitation, electro-oxidation and ion exchange 2 .
In the manuscript, our proposed evaporative separation is first-of-all targeted on the separation of metals from the matrices (such as plastic, ceramics and carbon,) of e-waste. Such a high temperature (~3000 K), and assisted by the additives, could evaporate most of the metals.
We did not seek to obtain the pure, individual metals. The evaporative separation scheme exhibits a capability for the separation of metals, which could be improved by further optimization of the FJH setup. We here first discuss the theoretical separation factor of metals based on the vapor pressure difference and the effect of alloy melt formation on the separation factors; second, we discuss the separation ability achieved now; third, we discuss the chemical additives assisted separation; fourth, we made evidence-based predictions regarding how to further improve the separation of the evaporative separation scheme.
Theoretical separation factors of the evaporative separation process based on the vapor pressure difference.
The e-waste contains most of the metals across the Periodic Table. Here, we choose the abundant metals in e-waste: Cu, Sn, Al, Fe, and Zn; the precious metals Rh, Pd, Pt, Ag, and Au; and the toxic heavy metals Hg, Cd, As, Pb, and Cr, as representatives to calculate their separation factors. The vapor pressure-temperature relationships of these metals and C are plotted in Supplementary Fig. 21a. If we do not consider the alloy effect on the vapor pressure, the theoretical separation factors of these metals could be calculated based on the vapor pressure differences by eq S10, where βA-B is the separation factor of metal A and metal B, In most cases, the evaporative separation has a large separation factor of >100, demonstrating that the evaporative separation is a potential process for metal separation. The heatmap shows that the metals could be grouped into a few clusters, the group with low boiling points: Hg, As, Cd, Zn, and Pb; the group with median boiling points: Ag, Al, Cu, Sn, Cr, Au, Fe, and Pd; and the group with high boiling points: Rh and Pt. The heavy metals tend to have low boiling points and are the easiest to be removed from the e-waste, followed by the abundant metal groups, and then the precious metal groups. For the elements with large vapor pressure difference, the separation factor is large, e.g., βAg-Pd ~ 2439; in contrast, the separation factor is small for the metals with similar evaporative behavior, e.g., βCu-Al ~ 1.25.

The effect of melt alloy formation on the separation factors.
There might be alloy formation during the FJH process, yet it is not a certainty. In e-waste, the metals are usually separated by the matrix substrates, and the heating rate is ultrafast (>10 4 K s -1 ) in the FJH process with short reaction duration (~1 s). As a result, the metals might not form alloys before their sublimation. Moreover, not all the metals will form a solid solution.
For example, Ag and Cu are thermodynamically immiscible, hence the melting will have no effect on their evaporative behaviors.
In the case of alloy melt formation, we discuss the partial vapor pressure of each metal components over alloys. For simplicity purposes, we consider the binary alloys AB. The equilibrium partial vapor pressure over liquid alloys is given by eq S11: where pA is the partial vapor pressure of metal A, 6 8 is the vapor pressure of pure metal A, and aA is the activity of component A in the alloy AB.
The activity has the following properties: (1) a = 1 for a pure phase that does not exhibit solid solution. Hence, for the metals that do not form alloys, like the case of Cu and Ag, the melting has no effect on their vapor pressure. (2) The activity is related to the mole fraction (x). In an ideal model, the activity equals to the mole fraction, 6 = 6 . In a non-ideal model, according to the normal activity-composition diagram (Supplementary Fig. 22), one component exhibits nearly ideal behavior at very low mole fractions or very high mole fractions. For such compositions, activity is approximately equal to mole fraction. In our cases, the precious metals and heavy metals are trace metals (~10 ppm level), and the major metals, such as Cu, Sn, and Al, are >1 wt% (>10000 ppm level) (Supplementary Fig. 14). Hence, the activity nearly equals the mole fraction for precious metals and heavy metals. In other words, the formation of solid melt will not have an apparent effect on their evaporative behavior. The The achieved separation ability by the evaporative separation.
The metal separation ability from the chemical additives.
In this manuscript, we demonstrate an improved recovery yield of precious metals by using the halide additives (Figs. 2a-f). The type of halide also affects the separation factors.  Tables 6-7). It is found that Cl works best for Ag, F works best for Rh and Pd, and I work best for Au.
Hence, the use of halides as additives could improve the recovery yield, while at the same time change the separation factors. The metal separation by the introduction of additives is attributed to the different chemical reactivity of precious metals with the chemical additives, which could be presumably further optimized for a better separation ability. For example, in future studies, we can first add Cl-containing additives to separate Ag, then F-containing additives to separate Rh and Pd, and then I-containing additives to separate Au. There is a tradeoff between recovery yield and separation ability; we focused on a high recovery yield in this manuscript and did not seek high separation ability.

The evidence-based predictions on the practices to increase the separation factors.
We noticed that the achieved separation factors are lower than the theoretically calculated ones (Supplementary Tables 3-7). To further increase the separation ability, we think that more carefully controlled temperatures and reaction duration would be helpful, which is the next step for the evaporative separation scheme. We are currently upgrading our FJH system with better temperature controllability. In the future, we presume to evaporate the metals oneby-one by progressively increasing the FJH temperature. We again note that there is always tradeoff between recovery yield and separation ability. Our current work mainly focused on the high recovery yield and hence put less effort toward the separation. Further work is essential to balance the recovery yield and metal separation. 13

Supplementary Note 5. The energy consumption and cost evaluation
The energy consumption was calculated using eq S12: Where E is the energy per gram (kJ g -1 ), V1 and V2 are the voltage before and after flash Joule heating, respectively, C is the capacitance (C = 60 mF), and M is the mass per batch. Smelter, Sweden used a Kaldo furnace for smelting 8 . The e-waste was converted into a mixed Cu alloy by the Kaldo furnace, which is similar to our collected solids by evaporative separation using the FJH setup. They reported the required energy of 274 GJ ton -1 for e-waste processing.
While in our case, the energy consumption is 3.38 GJ ton -1 , corresponding to ~1/80 of the Kaldo furnace. We note that the energy consumption is optimized for the commercial Kaldo furnace, and we presume the energy consumption of our FJH setup could be further reduced when scaling up to industrial scale.

Supplementary Note 6. Strategy for scaling up of the FJH process.
Joule heating is a mature heating technique and has been widely used in multiple practical devices and industrial processes, for example, electric fuses and electric heaters. The FJH disclosed here is intrinsically a Joule heating process. The difference of the FJH and conventional Joule heating technique lies in the modes of electrical energy supply and the heating duration. The conventional Joule heating process uses direct current (DC) or alternating current (AC) sources to provide a stable electric output. For our FJH process, a capacitor bank is used to provide a pulsed voltage output in a short time (down to ms scale). The FJH process is indeed highly scalable. Here, we first conduct the theoretical analysis on the scaling rule of the FJH process; second, we mention the batch-by-batch scaling up experiments with productivity up to kg scale in our research lab; third, we make an evidence-based prediction on how the FJH process could be scaled up by a continuous, roll-to-roll manner; fourth, we briefly discuss the undergoing industrial-scale application of the FJH process.
The scaling rule of FJH process revealed by theoretical analysis.
The accessible high temperature and the uniform temperature distribution across the sample are the two key points when scaling up the FJH process. For Joule heating, the heat amount (Q) is determined by eq S13, where I is the current passing through the sample, R is the resistance of the sample, and t is the discharging time. We then consider the heat per volume (Qv) determined by eq S14, where j is the current density, ρe is the electrical resistivity, and t is the discharging time.

15
The temperature is proportional to the heat since the heat capacity of the sample is constant.
Since the electrical resistivity of the sample is constant, to maintain a constant Qv and t when scaling up the sample, we need to maintain a constant j.
The charge (q) in the capacitor bank is determined by eq S15, where C is the total capacitance, and V is the charging voltage. If we suppose the charges in the capacitor bank are discharged in the discharging time of t, the current (I) passing through the sample is calculated by eq S16, Then, the current density (j) is determined by eq S17, where S is the cross-sectional area of the sample. Considering the cylinder-shaped sample (which is usually the case), the sample mass (m) is calculated by eq S18, where ρm is the density of the sample, S is the cross-sectional area of the sample, and L is the length of the sample.
Above all, we obtain a formula determining the current density of eq S19, As discussed, to scale up the sample mass (m), we need to maintain a constant current density.
This could be realized by the measures of (1) increasing the FJH voltage (V), and/or (2) increasing the capacitance (C).
In our FJH setup, we use a commercial aluminum electrolytic capacitor (450 V, 6 mF, Mouser #80-PEH200YX460BQU2). The state-of-art commercial aluminum electrolytic capacitor has the maximum rated voltage of V1 = 630 V, and capacitance of C1 = 2.7 F. In our typical experiment, we use a FJH voltage of V0 = 150 V and C0 = 0.06 F for the FJH of sample with mass of m0 = 0.2 g. According to eq S18, by using just one state-of-art capacitor, the mass is m1 = m0 (C1 V1)/(C0V0) = 37.8 g per batch. The capacitors could be connected in parallel to get a high total capacitance by eq S20, Considering that we use a capacitance bank composed of 30 aluminum capacitors with the total capacitance of Ctotal = 81 F, the mass will be mbatch = 1.1 kg per batch in the discharging time of 1 s. The re-charging of the capacitor bank is the slowest step of the process, which could be compensated by a high-speed charging technique. Supposing the total time is ttotal = 10 s per batch, one such FJH setup could process the e-waste of m ~ 10 tons per day. Based on our experience and these calculations, it is reasonable to conclude that the FJH process is highly scalable, with the capability for industry-scale application.
The demonstration of the scaling of the FJH process in our research lab.
In our typical experiment, we use a mass of m0 = 0.2 g, with the FJH condition of V0 = 150 V and C0 = 0.06 F. Here, we demonstrate the scaling up of FJH to a scale with mass of m1 = 2 g and m2 = 4 g per batch (Supplementary Fig. 23a). We used a capacitor bank composed of 104 aluminum capacitors (6 mF, 450 V, Mouser #80-PEH200YX460BQU2) in parallel, so the total capacitance is C1 = 0.624 F. For the sample mass of m1 = 2 g, we use a FJH voltage of V1 = 150 V, and for the sample mass of m1 = 2 g, we use a FJH voltage of V2 = 300 V, thus these conditions fit with the scaling rule of eq S19. Since temperature is a key parameter for the ewaste processing in our evaporative separation scheme, we recorded the temperature for those samples (Supplementary Figs. 23b-d). It is found that the temperature reaches >3000 K for all the samples, demonstrating the effective scaling up of the FJH process.

The evidence-based prediction of the continuous processing of e-waste by the FJH process.
In addition to the batch-by-batch process, we made evidence-based predictions for the continuous processing mode of the FJH processing of e-waste. As shown in Supplementary   Fig. 24, two baffles separate the feeds and the remaining solid. After the FJH, the bottom baffle is opened and the remaining solid is removed from the reactor. The top baffle is then opened, and the feeds are loaded into the reactor for the next FJH reaction. Since the collected volatiles in the cold trap are very little per FJH, it is not necessary to change the collection vessel. Note that this is only one possible continuous production method. Many mature engineering practices could be applied.

Industrial-scale application of the FJH process is underway.
The FJH process, which was invented by our group for the synthesis of flash graphene 9 , is already undergoing industrial-scale scaling up by Universal Matter, Inc The FJH process could achieve a high temperature, but the high-temperature region is limited to the sample. According to the Joule heating formula, = : , the heat amount is proportional to the resistance. The resistance values of the Cu electrode, the graphite electrode, and the sample are ~0.09 Ω, ~0.11 Ω, and >2.0 Ω, respectively. The Cu and graphite electrodes have much higher conductivity than the sample. Hence, the voltage drop was mainly imposed on the sample, and the heat amount generated by the discharging mostly retains on the sample.
During the FJH, the strong light emission region is limited to the sample (Supplementary Fig.   5a), indicating that Cu and graphite electrodes remain low temperature. The high thermal conductivity of the Cu electrodes also helps the fast thermal dissipation and prevents the high temperature. Moreover, the FJH time is very short, with the >3000 K temperature in tens of ms. The Cu electrode and graphite rod show no obvious change after the FJH other than the contamination of the Cu electrode by CB (Supplementary Fig. 5b). The resistance of the Cu 24 electrodes and the graphite electrodes remains the same after the FJH. Hence, the FJH process has no significant effect on the Cu and graphite electrodes. The precious metals as the late transition group metals usually have weak affinity with C and almost no solubility for carbon. The precious metals tend to not form carbide phases even at high temperature 12 . For example, there is no experimental evidence for a possible inorganic crystalline gold carbon compound. Experimentally, we mixed RhCl3, PdCl2, AgCl, and HAuCl4 with CB (5 wt% for each) and conducted the FJH. The XRD pattern of the product is shown in Supplementary Fig. 9. The XRD result showed that there were pure metal phases and metal alloy phases. No precious metal carbide phase was observed. Hence, the use of CB as conductive additives will not affect the evaporative behavior of precious metals. The mass ratios of halide in the raw materials were F, ~6.4%; Cl, ~2.3%; and I, ~3.7%. After FJH, the halide content in the remaining solids was F, ~3.8%; Cl, ~1.6%; and I, ~3.1%. This corresponds to the evaporative loss of F, ~40%; Cl, ~30%; and I, ~16%. The halide salts are expected to be easily recovered by a water washing and precipitation process due to their high solubility, while the components of the e-waste and carbon have low water solubility if any.

Supplementary
Hence, it is possible to recovery the halides either remaining in the solids or evaporated and collected in the cold trap. The use of the halides will not introduce significant additional materials cost. The element maps prove that the successful collection of precious metals. The metals spread over the entire collected solid.