pubs.acs.

org/JACS Article

Electric Field Effects in Flash Joule Heating Synthesis

Lucas Eddy,○ Shichen Xu,○ Changhao Liu, Phelecia Scotland, Weiyin Chen, Jacob L. Beckham,
Barbara Damasceno, Chi Hun Choi, Karla Silva, Alexander Lathem, Yimo Han, Boris I. Yakobson,
Xinfang Zhang,* Yufeng Zhao,* and James M. Tour*
Cite This: J. Am. Chem. Soc. 2024, 146, 16010−16019 Read Online

ACCESS Metrics & More Article Recommendations *

sı Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: Flash Joule heating has emerged as an ultrafast, scalable,

and versatile synthesis method for nanomaterials, such as graphene. Here,
Downloaded via CITY UNIV OF HONG KONG on April 9, 2025 at 04:10:15 (UTC).

we experimentally and theoretically deconvolute the contributions of

thermal and electrical processes to the synthesis of graphene by flash
Joule heating. While traditional methods of graphene synthesis involve
purely chemical or thermal driving forces, our results show that the
presence of charge and the resulting electric field in a graphene precursor
catalyze the formation of graphene. Furthermore, modulation of the
current or the pulse width affords the ability to control the three-step
phase transition of the material from amorphous carbon to turbostratic
graphene and finally to ordered (AB and ABC-stacked) graphene and
graphite. Finally, density functional theory simulations reveal that the
presence of a charge- and current-induced electric field inside the
graphene precursor facilitates phase transition by lowering the activation energy of the reaction. These results demonstrate that the
passage of electrical current through a solid sample can directly drive nanocrystal nucleation in flash Joule heating, an insight that
may inform future Joule heating or other electrical synthesis strategies.

■ INTRODUCTION
The application of electric fields in materials synthesis has
difference, typically up to several hundred volts, across a
reaction vessel that is several centimeters across, leading to
emerged as a prominent and captivating research area in average electric fields on the order of 103−104 V/m and a
modern science and engineering.1−3 An external electric field is current density of 105−107 A/m2. In the synthesis of flash
widely used to tune the activation energy of chemical graphene (FG), our previous studies have found that thermal
products,4 allowing thermochemical reactions to occur at annealing alone could not account for the formation of highly
lower temperatures,5 and to tune the crystallite size6 and turbostratic particles,12 suggesting a direct role for the passage
alignment7,8 of organic and inorganic products. The electric of current through the material in the nucleation of graphene
field strength used in these techniques is typically on the order crystals. Now that the synthesis of FG has been scaled up,16 a
of 104 to 109 V/m, though fields as low as 102 V/m have been deeper understanding of the FG growth process is necessary to
demonstrated to affect the reaction product.9 An external enable more precise control over the structure and quality,
electric field can change the synthesis products, and it has also namely, crystallinity, defect density, and turbostratic character,
been suggested that the presence of an electric field can have of FG for various applications.
an effect on the motion of electrically neutral particles.10 Here, we experimentally deconvolute the electrical and
Flash Joule heating (FJH), as a method exhibiting high- thermal effects of FJH in the conversion of metallurgical coke
energy efficiency and an ultrafast heating rate, has recently to graphene to examine the input that each has on the reaction
been used in the synthesis of many materials including enthalpy, activation energy, and product distribution. The
graphene,11−17 transition metal chalcogenides,18 inorganic enthalpy changes occurring during the FG synthesis using both
nanoparticles,19,20 and various other carbide and inorganic electrothermal and purely thermal flash reactions demonstrate
compounds.21−24 It has also been widely explored as a method
for effective remediation of soil25 and battery electrodes26 as
well as a method for recycling and upscaling materials.27−29 In Received: February 28, 2024
contrast to conventional chemical heating methods, which use Revised: April 24, 2024
conductive or radiative heating, Joule heating requires an Accepted: May 14, 2024
electric current to flow through the sample itself, which, in Published: May 28, 2024
turn, directly heats the feedstock. Passing this electric current
through the reactants involves imposing an electrical potential

© 2024 American Chemical Society https://doi.org/10.1021/jacs.4c02864

16010 J. Am. Chem. Soc. 2024, 146, 16010−16019
Journal of the American Chemical Society pubs.acs.org/JACS Article

Figure 1. Comparison between thermal and electrothermal systems on 120 mg MCAC. (a) Schematic illustration of the carbon-paper-based “2D”
flashing system used to perform thermal flashes. By this scheme, most electric current passes under rather than through the carbon feedstock. Thus,
the reaction mechanism is predominantly thermal conduction. (b) Schematic illustration of the standard, tube-based flashing system used to
perform electrothermal flashes. By this scheme, an electric current passes through the metallurgical coke amorphous carbon (MCAC) feedstock
itself, exposing it to an electric field and heating by resistive Joule heating between the granules of MCAC that are 0.30−0.84 mm in size. (c)
Temperature vs time profile of a thermal flash performed at 100 V. Additional discussion on the thermal heating of MCAC through thermal
conduction is given in Figure S1. (d) Representative Raman spectrum of the product of a thermal flash performed at 100 V. There is little graphene
character noted by the small and broad 2D peak. (e) Temperature vs time profile of an electrothermal flash performed at 80 V. The temperature
profiles of both reactions are comparable in both the max temperature achieved and the duration of the reaction. (f) Representative Raman
spectrum of the product of an electrothermal flash performed at 80 V. The product of this flash is predominantly FG, as demonstrated by the sharp
and high 2D peak, despite being reacted with less energy than in the thermal 100 V reaction. Additional Raman spectra and temperature profile
comparisons are illustrated in Figures S2−S4. (g) TEM image of MCAC reacted thermally at 100 V. The character is predominantly amorphous.
(h) XRD pattern of MCAC reacted thermally at 100 V. The width and low height of the (002) peak are indicative of low crystallinity. (i) TEM
image of MCFG formed electrothermally at 80 V. Clear graphene domains are visible. (j) XRD pattern of MCFG formed electrothermally at 80 V.
The sharp (002) peak indicates high crystallinity.

that the electric field strength affects the morphology of the blocking the current. First, we developed such a reaction vessel
carbon feedstock as it is converted from amorphous carbon (Figure 1a). During a flash reaction in this vessel, the electrical
(AC) to turbostratic (intersheet misalignment) FG and finally current will pass primarily through the carbon paper platform,
to ordered FG and graphite, which is composed of both Bernal which is more conductive than the metallurgical coke resting
(AB) and rhombohedral (ABC) stacked layers.30 In addition, above and in contact with the electrodes at both ends of the
both experimental and theoretical studies demonstrate that vessel; minimizing the current through the metallurgical coke.
oscillating the current density of the reaction through pulse In this way, the metallurgical coke is heated through thermal
width modulation facilitates this phase transition. Finally, conduction by its contact with the underlying carbon paper. By
theoretical simulations demonstrate that an induced electric contrast, the reaction tube architecture used in conventional
field inside the graphene precursor drives the phase transition flashing (Figure 1b) passes the electrical current through the
by decreasing the activation energy of the reaction. These metallurgical coke itself, which is in full contact with the
results show that chemical reactions can be driven directly by electrodes. These two distinct methods were performed with
the flow of current in flash Joule heating reactions, deepening samples with the same mass, resistance, and flashing energy.
our understanding of the mechanisms driving nanocrystal The thermal flash achieved ∼2000 °C (Figure 1c) and
formation in this process. produced graphene of poor quality (Figure 1d), resulting in

■ RESULTS AND DISCUSSION

Essential to the deconvolution of the electrical and thermal
similar flashing temperature profiles (Figure 1c,d). Despite
keeping all of these factors constant, the products produced by
these two reaction schemes are dramatically different.
effects of FJH is a reaction vessel architecture that permits the Raman spectroscopy is a commonly used technique for the
heat of a flash reaction to pass through the sample while identification and characterization of graphene. Raman spectra
16011 https://doi.org/10.1021/jacs.4c02864
J. Am. Chem. Soc. 2024, 146, 16010−16019
Journal of the American Chemical Society pubs.acs.org/JACS Article

Figure 2. Effect of electric field on the flash reaction and products. (a) Graphene quality comparison of MCAC flashed in 120 mg batches at the
same energy density (energy/mass) but using different flash architectures evaluated by Raman spectroscopy analysis. (b) Graphene conversion and
quality comparison of MCAC of comparable product quality at different flashing energy densities. The graphene yield of the product in the thermal
case is 9%, while the graphene yield of the product in the electrothermal case is 10%, which demonstrates the near-minimum energy requirement at
which conversion begins. Additional Raman spectra of electrothermal flashes at different energies are shown in Figures S6−S7. (c) Graphene
quality of MCAC flashed electrothermally at different electric fields but at a constant energy density. Higher electric fields were achieved by
increasing the voltage of the flash and reducing the capacitance such that the energy of each flash remained constant. The electric field strength was
determined by dividing the voltage of the flash by the length of the resistive sample through which the current flows. The graphene quality and
conversion are improved by the presence of a stronger electric field, even at a constant energy. The average Raman spectra of these scans are shown
in Figure S8. (d) Reaction progress diagram of activation energies by the thermal method at a 120 mg scale. *The activation energy from
turbostratic FG to graphite is estimated from the electrothermal data because we could not input enough energy to affect the thermal transition to
graphite. Specifically, comparing the first step measured energies of 799/200 = 4.00 and using the same ratio of difference: 410 × 4.00 = 1640. (e)
Reaction progress diagram of enthalpy by the electrothermal method. These results were obtained from flashing with a 120 mg batch size. The
activation energy to obtain turbostratic FG from MCAC was measured to be 200 kJ/mol (16.7 kJ/g), while the activation energy to obtain graphite
from turbostratic FG was measured to be 799 kJ/mol (66.5 kJ/g). (f) Density-functional theory (DFT)-determined reaction progress diagram
displaying simulated enthalpies of the transition reactants and products by a thermal heating scheme. (g) Reaction enthalpy of a previously reported
scaled (5.7 g) reaction.16 At this scale, the transition from MCAC to turbostratic FG is achieved at an activation energy of 90 kJ/mol (7.5 kJ/g).
The activation energy value for the turbostratic FG to graphite transition is extrapolated in the same manner as that in the thermal case.

of graphene and graphite contain the second-order overtone of resonance-enhanced turbostratic graphene.11,32 In a Raman
the D band, termed the two-dimensional (2D) band. The spectrum, graphene exhibits a G peak, which is common
appearance of this band is a positive indicator for the formation among ordered carbon materials, a D peak, which is correlated
of a graphitic lattice, and its shape and intensity offer important with the defect density, and a 2D peak, which changes in shape
nanostructural information.30,31 Spectra of high-quality FG and intensity in accordance with crystallinity and layer-
contain a resonance-enhanced single-Lorentzian 2D band with stacking. Consequently, we evaluate graphene conversion and
a narrow full width at half-maximum (fwhm; ∼16 cm−1) and a quality (crystallinity, defect density, and turbostratic character)
magnified I2D/IG peak intensity ratio (up to a maximum of 17) by considering the ratio of the intensity of the D peak to that
as compared to pristine single-layer graphene (which possesses of the G peak (termed ID/G) as a metric for defect density and
I2D/IG of ∼3). Statistics on individual spectra illustrate the ratio of the intensity 2D peak to that of the G peak (I2D/G)
relationships between the I2D/IG ratio, the fwhm of the 2D as a metric for graphene conversion. Even though a higher
band, and the position of the 2D band, showing a blue-shifted voltage, and consequently more energy, was used in the purely
and narrowed 2D band at high I2D/IG, which is indicative of thermal flash reaction, the graphene quality is significantly
16012 https://doi.org/10.1021/jacs.4c02864
J. Am. Chem. Soc. 2024, 146, 16010−16019
Journal of the American Chemical Society pubs.acs.org/JACS Article

Figure 3. Carbon phase transition process at different voltages. (a) TEM images of the products of 120 mg batches of MCAC flashed at 40 V, (b)
80 V, (c) 120 V, and (d) 160 V. As the MCAC feedstock is flashed at higher voltages, the products transition from an AC structure to a turbostratic
FG structure to finally a graphite structure. (e) Graphene quality of the products of these flashes as determined by Raman spectroscopy analysis.
Flashing at higher voltages improves graphene crystallinity and reduces defect density (increases the I2D/G and decreases the ID/G peak ratios). Only
points exhibiting an I2D/G ratio above 0.3 are assigned to be graphene. (f) XRD patterns highlighting the 002 peak at ∼26°, which indicates the
interlayer spacing of the carbon lattices. Higher voltages result in both a sharpening of this peak and a slight shift to higher angles, corresponding to
increased crystallinity and reduced interlayer spacing. (g) Illustration of the average interlayer spacing in nm above each bar of the flash products as
determined from the (002) XRD peak.

better in the electrothermal flash architecture, exhibiting a low has a dramatic effect on the graphene quality at a given energy.
ID/G and a high I2D/G (Figure 1e,f). A panel of flashes were performed at different energies under
The experimental results suggest that the passage of electric both experimental architectures until the graphene quality and
current itself through the metallurgical coke amorphous carbon conversion of a reaction with electrothermal architecture were
(MCAC) sample reduces the energy requirement for graphene comparable to the graphene quality and conversion of a
conversion. This finding is consistent with previous observa- reaction with thermal architecture (Figure 2b). Additional
tions of FG morphologies, in which two distinct morphologies reactions from this panel are shown in Figure S5. Even though
were found to form in FJH reactions using a carbon black ∼4.5× more energy (19 kJ/g) is used in the thermal
precursor.12 The “wrinkled” morphology was found to form architecture here, the graphene quality of the electrothermal
under simple thermal annealing and exhibited a Raman flash is superior. This is consistent with the idea that the
spectrum similar to the thermal-only flash shown in Figure passage of electrical current is necessary to drive the formation
1d. The turbostratic FG morphology exhibited large of large, turbostratic graphene flakes, which exhibit resonance-
turbostratic flakes and showed a Raman spectrum similar to enhanced I2D in the Raman spectrum.12 To explore this effect
that shown in Figure 1e. Thermal annealing alone was further, additional flash reactions were performed at constant
previously found to be insufficient to nucleate large energy and mass but at different flash voltages and therefore
turbostratic FG flakes. different electric field strengths (Figure 2c). These results
To examine this energy difference more precisely, we indicate that graphene conversion and quality are proportional
performed several panels of reactions under both the thermal to the strength of the electric field applied, while independent
and electrothermal reaction architectures and compared the of the energy of the reaction.
graphene quality of the products using Raman spectroscopy In theory, after conversion from MCAC to turbostratic
analysis. For each graphene product, Raman spectroscopy graphene by flash Joule heating, sufficient energy can further
measurements were performed twice on each of 100 points, convert turbostratic graphene to ordered graphene and
evenly spaced in a square grid across a 1 mm2 area of the graphite. Using a merely thermal reaction architecture, the
sample. From these 200 total scans, the average I2D/G and ID/G activation energy of the MCAC to turbostratic FG was
ratios were calculated. Among the measurements with a measured to be 410 kJ/mol (34 kJ/g) (Figure 2d).
discernible G peak, the portion of those with an I2D/G ratio of Electrothermal flashing results on the same scale are more
at least 0.3 was counted as a portion of the total of samples efficient than thermal flashing results (Figure 2e). The energy
with a discernible G peak. This portion is equal to the required to achieve conversion to ordered graphene by
graphene yield per an established protocol.17 electrothermal flash Joule heating at the same scale has been
When the flash reaction is performed with MCAC under measured here as ∼2× that required to achieve conversion to
both architectures at the same voltage and mass and thus at the turbostratic graphene. Density-functional theory (DFT)
same energy density (Figure 2a), the I2D/G peak ratio is higher calculations of thermal flash Joule heating-induced transition
and the ID/G peak ratio is lower in the electrothermal case. similarly predict that ∼2.5× more energy is required to convert
Furthermore, at this energy density of 6 kJ/g, the graphene turbostratic graphene to ordered graphene as is required to
yield is 98% in the electrothermal case but only 5% in the transition from AC to turbostratic graphene (Figure 2f).
thermal case. Hence, merely changing the flashing architecture Furthermore, these calculations reveal these two transitions to
16013 https://doi.org/10.1021/jacs.4c02864
J. Am. Chem. Soc. 2024, 146, 16010−16019
Journal of the American Chemical Society pubs.acs.org/JACS Article

Figure 4. Effect of temperature oscillation on the graphene phase transition. (a) Simulated energy of the structure while heating at a constant
temperature. The structure prior to heating is shown in Figure S18a. (b) Simulation illustration of AC after being heated for 40 ps at 3200 K. Some
initial ordering is visible in the structure. (c) Simulated energy of the structure while heating at a variable temperature over 100 ps. (d) Simulation
illustration of the resulting turbostratic graphene lattice after 100 ps at a variable temperature oscillating between 1000 and 4000 K at a rate of 1 K/
fs. Figure S25 illustrates the radial distribution function of the turbostratic graphene lattice relative to those of AC and graphite. (e) Experimental
current profiles of flashes on MCAC performed by PWM (red) and flashes performed via a direct current (DC) discharge (black). Both flashes
discharged the same amount of energy into the sample, but the current oscillates by a PWM at a frequency of 1 kHz. This results in a 1 kHz
oscillation in heating the sample. The inset illustrates the PWM duty cycles in closer detail. (f) Illustration of graphene quality of MCFG products
determined by Raman spectroscopy analysis.

be exothermic, suggesting that the release of chemical energy the region is determined to be graphite rather than merely
also assists the carbon feedstock in achieving the ∼3000 °C ordered graphene (Figures 3d and S9d).31 While nanoscale
afforded during the conversion of MCAC to high-yield FG. characterization methods can verify the existence of graphene
Larger-scale reactions suffer less heat loss to the environment in a sample, bulk characterization methods are essential to
per gram of graphene produced. Consequently, at 5.7 g scale, evaluate graphene conversion over a large sample. Raman
the activation energy of the MCAC to turbostratic FG spectroscopy likewise illustrates that as a function of the
transition is only 90 kJ/mol (7.5 kJ/g) as previously reported flashing voltage, the I2D/G peak ratio increases and the ID/G
(Figure 2g).16 This value is ∼2.6× less than that calculated for peak ratio decreases (Figures 3e and S10). As MCAC is
the same transition with a pure thermal reaction with 100% converted to turbostratic FG, turbostratic-indicative TS1 and
thermal efficiency. TS2 peaks begin to emerge (Figure S11a). As the turbostratic
The transition from MCAC to turbostratic graphene and FG is converted to ordered FG, these TS peaks vanish and a
finally to ordered graphene and graphite is accompanied by Bernal-indicative M peak emerges (Figure S11b,c). X-ray
microscopic and spectroscopic changes in the carbon product. diffractometry illustrates that the (002) graphene peak,
These changes can be observed by flashing the MCAC at corresponding to the lattice vector that is oriented
different voltages to stop the reaction at different points in perpendicular to the graphene sheets, sharpens and shifts to
transition (Figure 3). At sufficiently low voltage and therefore a higher angle as the carbon layers stack and decrease in
energy, MCAC remains unconverted and thus does not exhibit spacing (Figure 3f). As the turbostratic graphene layers align to
any ordered structure under transmission electron microscopy AB- or ABC-stacked orientations, the interlayer spacing
(TEM) (Figures 3a and S9a). As the flashing voltage increases, consequently decreases (Figures 3g and S12). Turbostratic
an ordered structure begins to emerge in the product along MC-derived FG typically exhibits an interlayer spacing of
with the alignment pattern indicative of turbostratic graphene ∼0.341 nm, while ordered graphene exhibits a spacing of 0.338
(Figures 3b,c and S9b,c). At high voltages, ordered graphitic nm. The disordered−ordered graphene ratio in the sample is
sheets are observed under TEM. When these sheets linearly proportional to the graphene interlayer spacing (Figure
accumulate to be 10 or more layers thick in a region, then S13). These interlayer spacing changes as well as the
16014 https://doi.org/10.1021/jacs.4c02864
J. Am. Chem. Soc. 2024, 146, 16010−16019
Journal of the American Chemical Society pubs.acs.org/JACS Article

Figure 5. Simulation results of the phase transition. (a) Simulation results illustrating the change in electric free energy during graphene sheet
alignment. The reaction from AC to FG is exothermic. (b) 2D simulation illustration of the charge density distribution across AC flakes when
subjected to an electric current. Electric charge accumulates on the AC surface, leading to heightened electric field strength localized around the
edges of the flakes. (c) Simulation results illustrating the change in electric free energy of graphene sheets when oriented at different angles relative
to four different electric current densities. The derivative of this graph is also shown in Figure S26. (d) Simulation results illustrating the effect of
graphene thickness along the c crystalline axis on the electric free energy when the sheets are aligned with respect to an electric current. The
thickness of the graphene lattice is proportional to the reduction in free energy induced by the electric current. This graph is also shown in more
detail in Figure S27.

emergence of a more prominent 2H (101) peak (Figure S14) energy barrier to lattice crystallization, the entropy is
provide more evidence of this transition in the bulk sample. simultaneously too high to permit crystallization. Thus,
As discussed previously, FG quality and conversion are a subsequent cooling phases are needed to facilitate lattice
function of the voltage, capacitance, energy, mass, and tube crystallization. Furthermore, the rate at which this transition
size.16,17 We also found that additional flashes, even at the occurs is proportional to both the frequency and the amplitude
same energy, can further improve graphene quality over that of the temperature oscillation (Figure S19). The effect that
produced by the initial flash (Figure S15). The improvement PWM has on graphene conversion is further demonstrated
in FG quality upon reflashing is proportional to the energy through experimental results. By introducing PWM duty
used (Figure S16). FG flashed at higher voltages and thus with cycles, an oscillating current profile can be obtained (Figure
different interlayer spacing has measurably different surface 4e). Despite being flashed at the same energy, FG synthesized
area and dispersibility in solution (Figure S17); hence, the by oscillating the current at 1 kHz with PWM exhibits slightly
ability to control the graphene characteristics is important. improved graphene quality (higher crystallinity and lower
Pulse-width modulation (PWM) has been previously defect density) and significantly improved graphene uniformity
demonstrated to improve both graphene yield and uniformity compared to FG synthesized by unmodulated direct current
when oscillated at 1 kHz frequency.16 Thus, the implementa- (DC) current (Figure 4f). Additional results of experiments
tion of oscillating the current by PWM, which, in turn, examining the effects of PWM on the graphene products are
oscillates the carbon heating profile, enables FG to be also illustrated in Figures S20−S22. Though PWM improves
produced at a lower energy cost. To investigate this further, graphene quality, small changes in the PWM frequency do not
DFT calculations were performed on the thermal heating have a significant effect on graphene quality (Figure S23).
synthesis of turbostratic graphene from AC (Figures 4, S18 and When reflashes are performed with more intense duty cycles,
S19), in one simulation while maintaining a constant the graphene quality is relatively unaffected, but the interlayer
temperature profile (Figure 4a,b) and in another while varying spacing decreases more significantly than that seen with less
the temperature at a constant frequency (Figure 4c,d). The intense duty cycles (Figure S24).
simulation of the constant temperature profile heating case Additional theoretical calculations were performed to further
demonstrates no phase transition from AC to graphene. elucidate the effect that running an electric current through the
However, when the temperature is allowed to oscillate, even at sample by an applied voltage has on flash Joule heating
the same average heating rate, a transition from AC to reaction (Figure 5). Under the influence of an electric field, the
turbostratic graphene is observed. This mechanism occurs randomly distributed carbon atoms in AC begin to align
because while a constant high temperature can break the themselves, ultimately forming aligned graphene layers with a
16015 https://doi.org/10.1021/jacs.4c02864
J. Am. Chem. Soc. 2024, 146, 16010−16019
Journal of the American Chemical Society pubs.acs.org/JACS Article

lower electric free energy (Figure 5a). As electric current converted to graphene as a portion of the total product. For the 200
passes through the carbon atoms, high current density and thus spectra of each sample, a custom Python program first determines the
a high localized electric field result on the surface of the carbon number of these spectra that have a discernible G peak. Among these
atoms (Figure 5b). Thus, even though the average electric field spectra, the program counts the portion of which has an I2D/G peak
ratio of at least 0.3. This portion is equal to the graphene yield. A
across a flashing vessel is ∼104 V/m, the electric field localized spectrum without a discernible G peak in a carbon sample most often
around these carbon atoms can be far higher. Additionally, the results from the Raman laser being insufficiently focused.
presence of this electric current energetically favors the X-ray Crystallography Analysis. X-ray diffractometry scans were
reorientation of the graphene layers along the direction of performed by using a Rigaku Smartlab II using zero background
the electric field (Figure 5c) and the ordering of the graphitic sample holders, a 40 kV, 30 mA X-ray source, and a Cu K-β filter.
lattice to reduce interlayer spacing (Figure 5d). Both of these Transmission Electron Microscopy Analysis. Transmission
effects are proportional to the strength of the applied voltage electron microscopy (TEM) images were collected using an FEI Titan
and therefore to that of the electric field present in the reaction Themis operating at 300 keV with a convergence angle of 25 mrad.
sample. The aliquot of sample (5 mg of sample/1 mL of ethanol) was
prepared and was sonicated for 1 h to ensure homogeneous

■ CONCLUSIONS
We have demonstrated through both experiment and
dispersion. The resultant aliquot was drop cast on a Cu/lacey carbon
TEM grid (Ted Pella). The resultant grid was dried at 80 °C with
overnight vacuum drying.
theoretical simulation that the conversion of AC to FG by
flash Joule heating is an electrothermal rather than merely a
thermal process, and the conversion process can be further
assisted by using a PWM to oscillate the current. When the
■ COMPUTATIONAL METHODS
Density Functional Theory (DFT). Density functional theory
(DFT) methods33 are used as they are implemented in the Vienna
electric current passes through the reactant itself, it can greatly Ab-initio Simulation Package (VASP).33 A plane wave expansion up
facilitate the phase transition process. By changing the flash to 500 eV is employed in combination with an all-electron-like
voltage, the phase control from AC to turbostratic graphene projector augmented wave (PAW) potential.34 Exchange correlation
and then to graphite can also be realized. Both experimental is treated within the generalized gradient approximation (GGA) using
data and DFT calculations show that the energy requirements the functional parametrized by Perdew−Burke−Ernzerhof.35 For all of
in the electrothermal process are ∼2× lower than in the the MD simulation of turbostratic graphitic materials, the supercells
thermal process. The energy for conversion from turbostratic are big enough (with 720 C atoms inside the box of size ∼18.0 Å ×
graphene to graphite is ∼4 times higher than conversion from 18.0 Å × 18.0 Å), and only the Γ point is used for the Brillouin zone
AC to turbostratic graphene. Finally, the finite difference integration over Monkhorst−Pack type mesh.36 The van der Waals
interaction is evaluated within the DFT-D3 method developed by
method was used to analyze and explain the significant roles of Grimme et al.37 In structure optimization using the conjugate-gradient
current and electric field in the phase transition process. algorithm, both the positions of atoms and the unit cells are fully

■ EXPERIMENTAL SECTION
Materials. Metallurgical coke was obtained from SunCoke in
relaxed so that the maximum force on each atom is smaller than 0.01
eV/Å. The MD simulation is performed using a Nose−Hoover
thermostat and NVT ensemble with a time step of 1 fs. For
chunks of ∼10 cm in diameter. These chunks were then ground to calibration, we optimized the crystal lattice structure of graphite, and
diameters of 0.30−0.84 mm. Additional characterization of this raw the standard lattice constants a = 2.467 Å and c = 6.794 Å were
material has been previously reported.16 obtained.
Flash Joule Heating. Flash Joule heating reactions were For modeling the growth of turbostratic graphene layers, we use
performed by using two separate flash Joule heating systems. The randomly generated coordinates of carbon atoms inside a cubic box
reactions done at variable electric fields presented in Figure 2c were with a C−C distance of no less than 1.5 Å. To mimic the high
performed using an unmodulated DC system with additional nonequilibrium FJH synthesis, a variable temperature MD (VTMD)
capacitor banks that were connected to the system such that flash simulation was used with the temperature fluctuating at the range of
reactions could be performed at any 18 mF interval between 59.2, and 1500−4000 K at a rate of 0.2−1.0 K/fs. The VTMD mode of
221.2 mF. All other flash reactions were performed using a 0.624 F modeling significantly facilitates the crystallization. In contrast, the
capacitor discharge system previously exhibited.16 Among the normal constant temperature MD (CTMD) followed by fast cooling
reactions performed with this system, only the DC reactions more likely leads to the formation of AC with a three-dimensional
presented in Figure 4 implemented an unmodulated DC discharge. (3D) bonding network. Therefore, CTMD was used to model the
All other reactions used a 1 kHz pulse-width-modulated discharge formation of AC, and the VTMD was used to simulate the growth of
with a duty cycle sequence, except when explicitly stated, of 10% for 1 multilayered turbostratic graphene. All of the MD simulation jobs
s, followed by 20% for 0.5 s, and ending at 50% for 5 s. were started from the same initial atomic configuration in which the
All flash Joule heating reactions are operated by using a custom coordinates of the C atoms were randomly generated. The time scale
LabVIEW program. Current profiles are measured at 50 μs resolution of simulation is set typically to be 50 ps for the CTMD of AC and 100
by a Tamura L34S1T2D15 Hall effect sensor, operated by a custom ps for the VTMD turbostratic graphene, which were followed by fast
LabVIEW program. Temperature is measured using a Micro-Epsilon cooling to 100 K at the rate of 0.2 K/fs. Then, the structures were
CTRM1H1SF100-C3 pyrometer at 1 ms resolution. optimized to obtain the total energy and material density. Regardless
Raman Spectroscopy Analysis. Raman spectra were collected of the structural difference in different situations, the values of energy
using a Renishaw inVia microscope outfitted with a 5 mW, 532 nm and material density of AC are consistently similar. So are those for
laser and a 50× objective lens. Except where explicitly stated, all turbostratic graphene. But, the AC is distinctively different from the
Raman spectra were collected as follows: 2 Raman spectra were turbostratic graphene.
measured on each of 100 points across a square 1 mm2 grid on the Phase transition from AC to layered turbostratic graphene was
sample using a map scan with LiveTrack focusing software enabled. regularly realized in the VTMD simulations; however, the transition
The average Raman spectrum shown for each sample is the average of from layered turbostratic graphene to an ideal, ordered graphene was
these 200 spectra. The I2D/G and ID/G ratios and their standard never realized within the similar time scale of simulation. This
deviations are calculated from these 200 spectra per sample as well indicates that the activation energy of the latter process is much
using a custom Python program as previously described.32 Graphene higher than that of the former. Using the Arrhenius equation, P =
yield is a metric that describes the amount of product that has been v0e−Ea/kT by setting P = 1/100 ps, T = 2500 K (average), and v0 = 1015

16016 https://doi.org/10.1021/jacs.4c02864
J. Am. Chem. Soc. 2024, 146, 16010−16019
Journal of the American Chemical Society pubs.acs.org/JACS Article

Hz, the estimated activation energy for the phase transition from AC James M. Tour − Department of Chemistry, Department of
to turbostratic graphene is Ea = 2.5 eV or 236 kJ/mol. The mechanism Materials Science and NanoEngineering, Smalley-Curl
of phase transition from turbostratic graphene to ideal graphene is the Institute, the NanoCarbon Center, and the Rice Advanced
annihilation of the topological defects such as the 5/7/7/5 through Materials Institute, and Department of Computer Science,
Stone−Wales bond flips.38,39 Therefore, the activation energy of
phase transition from turbostratic graphene to ideal graphene is
Rice University, Houston, Texas 77005, United States;
roughly that of the bond flips, ∼6.0 eV or 579 kJ/mol according to orcid.org/0000-0002-8479-9328; Email: [email protected]
Samsonidze et al.40
Finite Difference Method. To explore the phase transition under Authors
the influence of the current-induced electric field, a finite difference Lucas Eddy − Applied Physics Graduate Program and
method was used to calculate the electrical free energy. The first Smalley-Curl Institute, Rice University, Houston, Texas
model was divided into 200 × 200 grids, with each grid having a size 77005, United States; Department of Chemistry, Rice
of 0.1 × 0.1 nm2. Within these grids, circular shapes represented University, Houston, Texas 77005, United States
disordered carbon atoms, while rectangle shapes represented Shichen Xu − Department of Chemistry, Rice University,
graphene sheets formed by the fusion of carbon atoms. This allowed
Houston, Texas 77005, United States
us to calculate the change in electrical free energy as irregular carbon
atoms were orderly arranged into graphene sheets under the influence Changhao Liu − School of Metallurgical and Ecological
of the electric field, followed by the subsequent orderly arrangement Engineering, University of Science and Technology Beijing,
of graphene sheets. The electrical potential within this system satisfied Beijing 100083, P. R. China
Laplace equation: ∇2φc = 0 and ∇2φg = 0, where φc and φg are Phelecia Scotland − Department of Materials Science and
potentials of carbon atoms and graphene sheet, respectively. To NanoEngineering, Rice University, Houston, Texas 77005,
investigate the effect of the electric field, particularly the local electric United States
field, on the transition from AC to graphene, we further calculated the Weiyin Chen − Department of Chemistry, Rice University,
current density distribution when the current passes through Houston, Texas 77005, United States
randomly arranged carbon atoms in case one of the first model.
Jacob L. Beckham − Department of Chemistry, Rice
The electric current density (J)⃗ can be determined using Ohm’s law: J ⃗
= −σ·∇φ. Due to the poor electrical conductivity of AC, the charge University, Houston, Texas 77005, United States
mainly accumulates on the surface of AC. Under the effect of the Barbara Damasceno − Department of Chemistry, Rice
electric field, AC transforms into graphene sheets, then causing the University, Houston, Texas 77005, United States;
rotation of graphene sheets to align in the direction of the current. In orcid.org/0000-0002-6284-6939
the second model, a rectangular shape with a size of 1 × 20 nm2 Chi Hun Choi − Department of Materials Science and
represents the graphene sheet. There is an angle θ between the NanoEngineering, Rice University, Houston, Texas 77005,
graphene and the current direction. θ was set between 10 and 90°. United States
The changes in electric free energy of the graphene sheets were Karla Silva − Department of Chemistry, Rice University,
studied at an applied voltage of 40, 80, 120, and 160 V. To investigate
Houston, Texas 77005, United States
the changes in interlayer spacing of graphene sheets under the
influence of an electric field, a third model was established. In this Alexander Lathem − Applied Physics Graduate Program and
model, there are a total of 15 layers of graphene sheets, with each Smalley-Curl Institute, Rice University, Houston, Texas
graphene layer represented by a rectangular shape measuring 1 × 20 77005, United States; Department of Chemistry, Rice
nm2. The changes in electric free energy were calculated during the University, Houston, Texas 77005, United States
variation in interlayer spacing of the graphene layers under voltages of Yimo Han − Department of Materials Science and
40, 80, 120, and 160 V. Considering the experimental results, the NanoEngineering, Rice University, Houston, Texas 77005,
maximum interlayer spacing measured was 0.346 nm, while the United States; orcid.org/0000-0003-0563-4611
minimum was 0.338 nm. Therefore, we only considered the changes Boris I. Yakobson − Department of Chemistry, Department of
in electric potential energy for 15 layers of graphene within a
Materials Science and NanoEngineering, and Smalley-Curl
thickness range of 5.20 to 5.05 nm.
Institute, the NanoCarbon Center, and the Rice Advanced
■
*
ASSOCIATED CONTENT
sı Supporting Information
Materials Institute, Rice University, Houston, Texas 77005,
United States; orcid.org/0000-0001-8369-3567
The Supporting Information is available free of charge at Complete contact information is available at:
https://pubs.acs.org/doi/10.1021/jacs.4c02864. https://pubs.acs.org/10.1021/jacs.4c02864
Additional graphs; a photograph of the experimental
Author Contributions
setup; Raman spectra; temperature profiles; TEM ○
images; XRD patterns; BET data; and graphs and L.E. and S.X. contributed equally to this work.
illustrations of simulation results (PDF) Notes

■ AUTHOR INFORMATION
Corresponding Authors
The authors declare the following competing financial
interest(s): Rice University owns intellectual property on the
Xinfang Zhang − School of Metallurgical and Ecological flash Joule heating strategy. J. M. T. is a stockholder in
Engineering, University of Science and Technology Beijing, Universal Matter Ltd., a company that has licensed the
Beijing 100083, P. R. China; Email: xfzhang@
intellectual property from Rice University. J. M. T. is not an
alum.imr.ac.cn
Yufeng Zhao − Department of Materials Science and officer, director or employee of Universal Matter Ltd. All
NanoEngineering, Rice University, Houston, Texas 77005, conflicts of interest are managed through regular disclosure to
United States; Department of Physics, Corban University,
Salem, Oregon 97317, United States; Email: yzhao@ the Rice University Office of Sponsored Programs and
corban.edu Research Compliance.
16017 https://doi.org/10.1021/jacs.4c02864
J. Am. Chem. Soc. 2024, 146, 16010−16019
Journal of the American Chemical Society pubs.acs.org/JACS Article

■ ACKNOWLEDGMENTS
The funding of the research was provided by the Air Force
(16) Eddy, L. J.; Luong, D. X.; Beckham, J. L.; Wyss, K. M.;
Cooksey, T.; Scotland, P.; Choi, C. H.; Chen, W.; Advincula, P.;
Zhang, Z.; Macevski, V.; Tour, J. Automated Laboratory Kilogram-
Office of Scientific Research (FA9550-22-1-0526, J.M.T.), the Scale Graphene Production from Coal. Small Methods 2023, 8,
Department of Energy, NETL (DE-FE0031794, J.M.T.), and No. 2301144, DOI: 10.1002/smtd.202301144.
the U.S. Army Corps of Engineers, ERDC (W912HZ-21-2- (17) Sattari, K.; Eddy, L.; Beckham, J. L.; Wyss, K. M.; Byfield, R.;
0050, J.M.T., B.I.Y., and Y.Z.). The characterization equipment Qian, L.; Tour, J. M.; Lin, J. A Scientific Machine Learning
used in this project was partly from the Shared Equipment Framework to Understand Flash Graphene Synthesis. Digital
Authority (SEA) at Rice University. Discovery 2023, 2, 1209−1218.
(18) Chen, W.; Wang, Z.; Bets, K. V.; Luong, D. X.; Ren, M.;

■ REFERENCES
(1) Huang, P.; Zhu, R.; Zhang, X.; Zhang, W. Effect of Free Radicals
Stanford, M. G.; McHugh, E. A.; Algozeeb, W. A.; Guo, H.; Gao, G.;
Deng, B.; Chen, J.; Li, J. T.; Carsten, W. T.; Yakobson, B. I.; Tour, J.
M. Millisecond Conversion of Metastable 2d Materials by Flash Joule
and Electric Field on Preparation of Coal Pitch-Derived Graphene Heating. ACS Nano 2021, 15, 1282−1290.
Using Flash Joule Heating. J. Chem. Eng. 2022, 450, No. 137999. (19) Yao, Y.; Huang, Z.; Xie, P.; Lacey, S. D.; Jacob, R. J.; Xie, H.;
(2) Wang, J.; Park, J.-H.; Lu, A.-Y.; Kong, J. Electrical Control of Chen, F.; Nie, A.; Pu, T.; Rehwoldt, M.; Yu, D.; Zachariah, M. R.;
Chemical Vapor Deposition of Graphene. J. Am. Chem. Soc. 2022, Wang, C.; Shahbazian-Yassar, R.; Li, J.; Hu, L. Carbothermal Shock
144, 22925−22932. Synthesis of High-Entropy-Alloy Nanoparticles. Science 2018, 359,
(3) Xu, S.; Wang, S.; Chen, Z.; Sun, Y.; Gao, Z.; Zhang, H.; Zhang, J. 1489−1494.
Electric-Field-Assisted Growth of Vertical Graphene Arrays and the (20) Chen, Y.; Egan, G. C.; Wan, J.; Zhu, S.; Jacob, R. J.; Zhou, W.;
Application in Thermal Interface Materials. Adv. Funct. Mater. 2020, Dai, J.; Wang, Y.; Danner, V. A.; Yao, Y.; Fu, K.; Wang, Y.; Bao, W.;
30, No. 2003302. Li, T.; Zachariah, M. R.; Hu, L. Ultra-Fast Self-Assembly and
(4) Wan, M.; Yue, H.; Notarangelo, J.; Liu, H.; Che, F. Deep Stabilization of Reactive Nanoparticles in Reduced Graphene Oxide
Learning-Assisted Investigation of Electric Field−Dipole Effects on Films. Nat. Commun. 2016, 7, No. 12332, DOI: 10.1038/
Catalytic Ammonia Synthesis. JACS Au 2022, 2, 1338−1349. ncomms12332.
(5) Jesus, L. M.; Silva, R. S.; Raj, R.; M’Peko, J.-C. Electric Field- (21) Wang, C.; Ping, W.; Bai, Q.; Cui, H.; Hensleigh, R.; Wang, R.;
Assisted Ultrafast Synthesis of Nanopowders: A Novel and Cost- Brozena, A. H.; Xu, Z.; Dai, J.; Pei, Y.; Zheng, C.; Pastel, G.; Gao, J.;
Efficient Approach. RSC Adv. 2016, 6, 107208−107213. Wang, X.; Wang, H.; Zhao, J.-C.; Yang, B.; Zheng, X.; Luo, J.; Mo, Y.;
(6) Morris, P. D.; McPherson, I. J.; Edwards, M. A.; Kashtiban, R. J.; Dunn, B.; Hu, L. A General Method to Synthesize and Sinter Bulk
Walton, R. I.; Unwin, P. R. Electric Field-Controlled Synthesis and Ceramics in Seconds. Science 2020, 368, 521−526.
Characterisation of Single Metal−Organic-Framework (MOF) Nano- (22) Chen, G.; Wang, Y.; Wang, X.; Zhao, Y.; Dong, Q.; Hao, L.;
particles. Angew. Chem., Int. Ed. 2020, 59, 19696−19701. Hong, M.; Guo, M.; Qiao, H.; Xiong, W.; Hu, L. Target-Sintering of
(7) Qin, R. Electric-Field-Induced Alignment of Electrically Neutral Single-Phase Bulk Intermetallics Via a Fast-Heating-Induced Rapid
Disk-Like Particles: Modelling and Calculation. Sci. Rep. 2017, 7, Interdiffusion Mechanism. ACS Mater. Lett. 2022, 4, 480−486.
No. 8449. (23) Dong, Q.; Hong, M.; Gao, J.; Li, T.; Cui, M.; Li, S.; Qiao, H.;
(8) Cheng, Y.; Cui, G.; Liu, C.; Liu, Z.; Yan, L.; Liu, B.; Yuan, H.; Brozena, A. H.; Yao, Y.; Wang, X.; Chen, G.; Luo, J.; Hu, L. Rapid
Shi, P.; Jiang, J.; Huang, K.; Wang, K.; Cheng, S.; Li, J.; Gao, P.;
Synthesis of High-Entropy Oxide Microparticles. Small 2022, 18,
Zhang, X.; Qi, Y.; Liu, Z. Electric Current Aligning Component Units
No. 2104761.
During Graphene Fiber Joule Heating. Adv. Funct. Mater. 2022, 32,
(24) Chen, W.; Li, J. T.; Ge, C.; Yuan, Z.; Algozeeb, W. A.;
No. 2103493.
Advincula, P. A.; Gao, G.; Chen, J.; Ling, K.; Choi, C. H.; McHugh, E.
(9) Palyanov, Y. N.; Borzdov, Y. M.; Sokol, A. G.; Bataleva, Y. V.;
A.; Wyss, K. M.; Luong, D. X.; Wang, Z.; Han, Y.; Tour, J. M.
Kupriyanov, I. N.; Reutsky, V. N.; Wiedenbeck, M.; Sobolev, N. V.
Turbostratic Boron−Carbon−Nitrogen and Boron Nitride by Flash
Diamond Formation in an Electric Field under Deep Earth
Conditions. Sci. Adv. 2021, 7, No. eabb4644. Joule Heating. Adv. Mater. 2022, 34, No. 2202666, DOI: 10.1002/
(10) Zhang, X.; Qin, R. Electric Current-Driven Migration of adma.202202666.
Electrically Neutral Particles in Liquids. Appl. Phys. Lett. 2014, 104, (25) Deng, B.; Carter, R. A.; Cheng, Y.; Liu, Y.; Eddy, L.; Wyss, K.
No. 114106. M.; Ucak-Astarlioglu, M. G.; Luong, D. X.; Gao, X.; JeBailey, K.;
(11) Luong, D. X.; Bets, K. V.; Algozeeb, W. A.; Stanford, M. G.; Kittrell, C.; Xu, S.; Jana, D.; Torres, M. A.; Braam, J.; Tour, J. M.
Kittrell, C.; Chen, W.; Salvatierra, R. V.; Ren, M.; McHugh, E. A.; High-Temperature Electrothermal Remediation of Multi-Pollutants in
Advincula, P. A.; Wang, Z.; Bhatt, M.; Guo, H.; Mancevski, V.; Soil. Nat. Commun. 2023, 14, No. 6371.
Shahsavari, R.; Yakobson, B. I.; Tour, J. M. Gram-Scale Bottom-up (26) Chen, W.; Salvatierra, R. V.; Li, J. T.; Kittrell, C.; Beckham, J.
Flash Graphene Synthesis. Nature 2020, 577, 647−651. L.; Wyss, K. M.; La, N.; Savas, P. E.; Ge, C.; Advincula, P. A.;
(12) Stanford, M. G.; Bets, K. V.; Luong, D. X.; Advincula, P. A.; Scotland, P.; Eddy, L.; Deng, B.; Yuan, Z.; Tour, J. M. Flash Recycling
Chen, W.; Li, J. T.; Wang, Z.; McHugh, E. A.; Algozeeb, W. A.; of Graphite Anodes. Adv. Mater. 2023, 35, No. 2207303.
Yakobson, B. I.; Tour, J. M. Flash Graphene Morphologies. ACS Nano (27) Deng, B.; Luong, D. X.; Wang, Z.; Kittrell, C.; McHugh, E. A.;
2020, 14, 13691−13699. Tour, J. M. Urban Mining by Flash Joule Heating. Nat. Commun.
(13) Wyss, K. M.; Beckham, J. L.; Chen, W.; Luong, D. X.; Hundi, 2021, 12, No. 5794.
P.; Raghuraman, S.; Shahsavari, R.; Tour, J. M. Converting Plastic (28) Deng, B.; Meng, W.; Advincula, P. A.; Eddy, L.; Ucak-
Waste Pyrolysis Ash into Flash Graphene. Carbon 2021, 174, 430− Astarlioglu, M. G.; Wyss, K. M.; Chen, W.; Carter, R. A.; Li, G.;
438. Cheng, Y.; Nagarajaiah, S.; Tour, J. M. Heavy Metal Removal from
(14) Chen, W.; Ge, C.; Li, J. T.; Beckham, J. L.; Yuan, Z.; Wyss, K. Coal Fly Ash for Low Carbon Footprint Cement. Commun. Eng.
M.; Advincula, P. A.; Eddy, L.; Kittrell, C.; Chen, J.; Luong, D. X.; 2023, 2, No. 13, DOI: 10.1038/s44172-023-00062-7.
Carter, R. A.; Tour, J. M. Heteroatom-Doped Flash Graphene. ACS (29) Chen, W.; Chen, J.; Bets, K. V.; Salvatierra, R. V.; Wyss, K. M.;
Nano 2022, 16, 6646−6656. Gao, G.; Choi, C. H.; Deng, B.; Wang, X.; Li, J. T.; Kittrell, C.; La, N.;
(15) Advincula, P. A.; Meng, W.; Eddy, L. J.; Beckham, J. L.; Eddy, L.; Scotland, P.; Cheng, Y.; Xu, S.; Li, B.; Tomson, M. B.; Han,
Siqueira, I. R.; Luong, D. X.; Chen, W.; Pasquali, M.; Nagarajaiah, S.; Y.; Yakobson, B. I.; Tour, J. M. Battery Metal Recycling by Flash Joule
Tour, J. M. Ultra-High Loading of Coal-Derived Flash Graphene Heating. Sci. Adv. 2023, 9, No. eadh5131.
Additives in Epoxy Composites. Macromol. Mater. Eng. 2023, 308, (30) Pimenta, M. A.; Dresselhaus, G.; Dresselhaus, M. S.; Cançado,
No. 2200640. L. G.; Jorio, A.; Saito, R. Studying Disorder in Graphite-Based

16018 https://doi.org/10.1021/jacs.4c02864
J. Am. Chem. Soc. 2024, 146, 16010−16019
Journal of the American Chemical Society pubs.acs.org/JACS Article

Systems by Raman Spectroscopy. Phys. Chem. Chem. Phys. 2007, 9,

1276−1290.
(31) Dresselhaus, M. S.; Jorio, A.; Filho, A. G. S.; Saito, R. Defect
Characterization in Graphene and Carbon Nanotubes Using Raman
Spectroscopy. Philos. Trans. R. Soc., A 2010, 368, 5355−5377,
DOI: 10.1098/rsta.2010.0213.
(32) Beckham, J. L.; Wyss, K. M.; Xie, Y.; McHugh, E. A.; Li, J. T.;
Advincula, P. A.; Chen, W.; Lin, J.; Tour, J. M. Machine Learning
Guided Synthesis of Flash Graphene. Adv. Mater. 2022, 34,
No. 2106506.
(33) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.;
Sutton, A. P. Electron-Energy-Loss Spectra and the Structural Stability
of Nickel Oxide: An LSDA+U Study. Phys. Rev. B 1998, 57, No. 1505,
DOI: 10.1103/PhysRevB.57.1505.
(34) Blöchl, P. E. Projector Augmented-Wave Method. Phys. Rev. B
1994, 50, No. 17953, DOI: 10.1103/PhysRevB.50.17953.
(35) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient
Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868.
(36) Monkhorst, H. J.; Pack, J. D. Special Points for Brillouin-Zone
Integrations. Phys. Rev. B 1976, 13, No. 5188, DOI: 10.1103/
PhysRevB.13.5188.
(37) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and
Accurate Ab Initio Parametrization of Density Functional Dispersion
Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010,
132, No. 154104.
(38) Zhao, Y.; Yakobson, B. I.; Smalley, R. E. Dynamic Topology of
Fullerene Coalescence. Phys. Rev. Lett. 2002, 88, No. 185501.
(39) Zhao, Y.; Lin, Y.; Yakobson, B. I. Fullerene Shape Trans-
formations Via Stone-Wales Bond Rotations. Phys. Rev. B 2003, 68,
No. 233403.
(40) Samsonidze, G. G.; Samsonidze, G. G.; Yakobson, B. I. Kinetic
Theory of Symmetry-Dependent Strength in Carbon Nanotubes.
Phys. Rev. Lett. 2002, 88, No. 065501.

16019 https://doi.org/10.1021/jacs.4c02864
J. Am. Chem. Soc. 2024, 146, 16010−16019