Methane pyrolysis by Joule heating for graphitic carbon and hydrogen production

Summary

The global energy transition toward sustainability requires technologies that can decarbonize energy carriers and fuels while producing valuable materials. Methane, a primary component of natural gas, is both a high-energy-density fuel and a significant greenhouse gas. This study reports an approach for methane pyrolysis utilizing Joule heating within the deposition substrate to drive the endothermic reaction. With electric current passing through a resistive porous carbon cloth, heat is generated to break C-H bonds of methane molecules. The decomposition of methane as it flows through the cloth results in hydrogen production and the formation of conformally layered graphite around the carbon fibers. The effects of input power, chamber pressure, feedstock flow rate, and process duration on hydrogen and graphite production are characterized via in situ mass spectrometry and laser absorption spectroscopy, resulting in methane conversion rates up to 88%, with hydrogen and carbon yields of 82% and 72%, respectively. Material characterization verifies uniform high-quality graphite deposition, with a Raman I_D/I_G ratio of 0.1 and 3.38 Å d-spacing. This Joule heating method for catalyst-free methane pyrolysis offers the potential for advancing hydrogen production technology by simultaneously producing valuable materials such as solid graphite, thus enhancing the economic viability of the fuel decarbonization process.

Graphical abstract

Highlights

• •Catalyst-free methane pyrolysis via Joule-heated carbon cloth demonstrated
• •Simultaneous hydrogen and high-quality graphite co-production achieved
• •Methane conversion rates reached up to 88% with hydrogen and carbon yields of 82% and 72%
• •Low Raman I_D/I_G ratio (0.1) and 3.38 Å graphitic interlayer spacing achieved

Energy application; Energy Resources; Thermodynamics

Introduction

Methane, a primary constituent of natural and shale gas, has a high hydrogen-to-carbon ratio, resulting in the highest energy density among hydrocarbons. This characteristic has led to its widespread use in various industrial processes, electricity generation, and daily life.¹ Nevertheless, methane is also the second-most abundant greenhouse gas after carbon dioxide on Earth, contributing approximately 1,625% of the greenhouse gas effects.^2,3 Advances in methane conversion technologies for hydrogen production with concurrent carbon capture^4,5 incorporating operational concepts, reactor design, system optimization, and storage solutions are crucial for providing sustainable solutions⁶ to environmental challenges. These processes not only yield clean hydrogen fuel but also can produce high-value carbon materials.⁷

Methane pyrolysis involves an endothermic reaction without CO₂ emissions for hydrogen and solid carbon production, which follows the global dissociation reaction^8,9

$${\text{CH}}_4\to \mathrm{C}+2{\mathrm{H}}_2\left(\Delta H^{°}=74.6\text{kJ}/\text{mol}\right)$$ (Equation 1)

Typically occurring at temperatures exceeding 1,000 K, methane pyrolysis follows a series of intermediate reactions. Along these pathways, various light and stable hydrocarbons, such as acetylene, ethylene, and ethane, emerge as minor by-products.^9,10 Previous studies have explored methane pyrolysis using various heating techniques, utilizing temperatures as high as 2,400 K without catalysts.¹¹ Some studies have successfully lowered this temperature to 923 K by employing catalysts.^12,13 Others have explored electrochemical methane splitting in molten salts,¹⁴ where renewable energy sources such as photovoltaics can drive the process.¹⁵ Table 1 presents a summary of the key findings of methane pyrolysis from previous investigations classified by reactor type.

Table 1.

Summary of diverse methodologies in methane pyrolysis from prior studies

Reactor type	Catalyst	Power input	T (K)	tres (ms)	Conversion	Carbon type	Reference
Electric furnace	Ni-Cua,b,f,Alb	1.8 kW	923–998	–	0.12–0.48b, 0.62a	CB	Pinilla et al.12,13
Electric furnace	Ni-Fe/Fe oreb	–	973–1,163/3,273	–	0.38–0.61	Graphite	Pan et al.7; Pudukudy et al.16
Electric furnace	Ni-Pt-Pdc	3.4 kW	1,273–1,338	–	0.38–0.95	CB	Upham et al.17
Microwave	CBb	0.4–1 kW/2.45 GHz	1,023–1,489	5,000	0.30–0.90	CB	Dadsetan et al.18
DC plasma	Noned	12.6–16.7 kW	1,500–2,000	∼100	0.991–0.994	CB	Fincke et al.19
DC plasma	Nonee	52 kW	870–2,400	∼5,000–30,000	0.87	CB	Mašláni et al.11
AC plasma	Nonef	25–35 kW	1,943–1,983	–	0.995–0.996	CB	Fulcheri et al.20
Indirect solar	Noneg	2–9 kW	1,500–2,000	60–120	0.05–0.7	CB	Dahl et al.21
Indirect solar	Noneg	23–37 kW	1,608–1,928	37–71	0.72–1	CB	Rodat et al.22
Indirect solar	CBh	1 kWabs	1,273–1,473	120	0.58–0.98	CB	Abanades et al.23
Direct solar	CBi	<4.9 kW	1,300–1,600	1,100–2,400	0.42–0.99	CB	Maag et al.24
Direct solar	CB & metala	3,000 kW/m2	923–1,223	–	0.43–0.67	CB	Pinilla et al.25
Direct solar	Nonej	0.92–2.49 kW	1,370–1,750	2–530	0.22–0.96	Graphite	Abuseada et al.9,26

CB, Carbon Black.

^aGraphite reactor

^bPacked bed

^cFlow seeded

^dRotary bed

^eCarbon cloth

^fFluidized bed

^gBubble column

^hCarbon reactor

ⁱCeramic reactor

^jGraphite electrodes and stainless steel vessel.

Electric furnace

Electric furnaces^12,13 are commonly used in methane pyrolysis and provide a viable alternative to traditional fired reactors by employing electric energy. Pinilla et al.¹² investigated methane thermo-catalytic decomposition using Ni-Cu-Al catalysts in a fluidized bed reactor (FBR), achieving 60 L/h of hydrogen and 15 g/h of carbon nanofibers semi-continuously. In another study,¹³ they compared a rotary bed reactor (RBR) with an FBR, finding higher hydrogen yields and better catalyst longevity in the RBR using a nickel-copper catalyst. Pudukudy et al.¹⁶ synthesized nickel and iron catalysts via solid-state citrate fusion, demonstrating high activity and stability up to 900°C with a maximum hydrogen yield of 67%. Post-processing techniques, such as thermal treatment at temperatures up to 3,000°C or electrochemical methods, are often essential for producing and purifying graphitic carbon from methane pyrolysis. This graphitic carbon is used as both cathode and anode in high-performance dual-carbon batteries.⁷ Hu et al. used Joule heating for post-processing carbon black particles in a graphite “boat” inside a vacuum furnace to graphitize them²⁷ and reported moderate graphitization. Upham et al.¹⁷ explored molten metal alloy catalysts such as 27% nickel-73% bismuth, achieving 95% methane conversion to pure hydrogen at 1,065°C in a 1.1-m bubble column, offering a solution to solid catalyst deactivation. In molten alloy systems, the carbon floats to the surface, making it continuously removable. Wismann et al. proposed an approach that integrates electrically heated catalytic structures²⁸ directly into a steam-methane reforming reactor for hydrogen production. This integration ensures that the heat source is in close proximity to the reaction site, driving the reaction toward thermal equilibrium. This method can reduce reactor sizes by up to 100-fold,²⁸ improving thermal balance, enhancing catalyst efficiency, and reducing by-product formation.

Microwave and plasma

Various other electromagnetic techniques have been employed in methane pyrolysis to meet the high heating demand while upscaling the process. Dadsetan et al.¹⁸ introduced a non-plasma method using microwaves to decompose methane into hydrogen and solid carbon. This process involves carbon particles in a fluidized bed absorbing microwave energy, creating a hot medium (>1,200°C) that decomposes flowing methane and achieving over 90% hydrogen selectivity after about 500 h of cumulative experiments. Fincke et al.¹⁹ modified a DC plasma reactor, originally designed for methane-to-acetylene conversion, to increase carbon yield to 30%, while decreasing acetylene production. The specific energy requirement (SER) input for efficient gas-phase conversion of methane to hydrogen and solid carbon needs to be about 30% higher than the theoretical minimum (0.933 kWh/Nm³ H₂). The PlasGas plasma reactor, studied by Mašláni et al.¹¹ features a DC plasma torch stabilized by a water vortex and achieved methane conversions of 60%–88% at various flow rates (100–500 slm), producing hydrogen, methane, and solid carbon consisting of spherical particles around 1 mm in size. The SER in this work ranged from 5.2 to 1.38 kWh/Nm³ H₂ for flow rates of 100–500 slm, respectively. Fulcheri et al.²⁰ developed a thermal plasma process achieving over 99% methane conversion with hydrogen selectivity above 95%, offering potential energy savings at 25 kWh per kg of hydrogen compared to water electrolysis at 60 kWh per kg. Nevertheless, plasma torches operate at high power inputs (60 kW¹⁹–120 kW¹¹) and temperatures reaching up to 11,900 K, presenting challenges due to high energy density and ultrahigh temperatures.

Indirect solar heating

Several studies have reported indirect solar-based methods for converting methane into hydrogen and carbon black. Dahl et al.²¹ investigated the use of a solar furnace to heat an aerosol flow in a graphite tubular reactor, achieving a 70% dissociation of pure methane within residence times under 0.1 s and producing 20–40 nm amorphous carbon black particles. The potential for process scale-up could extend to a plant capable of producing 5 M kg/year of carbon black and 1.67 M kg/year of H₂. However, challenges in achieving complete dissociation were noted due to heat transfer limitations. Rodat et al.²² developed a pilot-scale 1-MW solar furnace with seven tubular zones and a graphite cavity-type solar receiver, achieving an 88% yield of hydrogen (200 g/h), 49% yield of carbon black (330 g/h), and 340 g/h of acetylene at 1,800 K, with a thermal efficiency of 15%. Abanades et al.²³ explored CO₂-free hydrogen production in a packed-bed reactor through solar thermo-catalytic decomposition, supporting continuous catalyst injection and replacement for scalable solutions. The study achieved up to 98% methane conversion to hydrogen at 1,200°C with a residence time of 0.12 s, producing acetylene as a secondary by-product.

Direct solar heating

For direct solar utilization, researchers have reported various approaches to methane decomposition. Maag et al.²⁴ investigated thermal methane decomposition using a 5-kW particle-flow solar chemical reactor in a solar furnace with carbon black seed particles. This method achieved over 95% methane conversion and high hydrogen yield within a short residence time of 2 s, with a solar-to-chemical energy conversion efficiency of 16%. The process led to the formation of filamentous agglomerations of product carbon on seed particles. Pinilla et al.²⁵ focused on solar catalytic methane decomposition using different catalysts. Metallic catalysts produced nanostructured carbons such as carbon nanofibers or multiwalled carbon nanotubes, while a carbonaceous catalyst generated amorphous carbon. These catalysts facilitated methane decomposition at lower temperatures (600°C–950°C), resulting in higher reaction rates, enhanced process efficiency, 100% selectivity to H₂, and improved quality of carbonaceous products compared to high-temperature non-catalytic solar methane decomposition. Abuseada et al.^9,26 developed a direct methane pyrolysis method employing a 10-kWe solar simulator and fibrous carbon starting substrate. This approach achieved methane conversion rates ranging from 22% to 96%, demonstrating significant production of hydrogen and graphitic carbon. A methane inlet flow of 2,000 sccm resulted in hydrogen production rates and solar-to-chemical efficiencies of up to 4.88 g/h and 3.05%, respectively. Despite the promising results, particularly in graphitic carbon production, further efforts are expected to optimize these processes and enhance the quality and efficacy of graphitic carbon for lithium-ion batteries.

The products from most of the above-mentioned methane pyrolysis approaches include hydrogen, other hydrocarbons, and carbon black. While some methods have produced graphitic solids as a by-product, they often require catalytic or post-heat/electrochemical treatments for refinement.⁷ Additionally, the quality of the graphite produced remains an area for further improvement.⁹

Here, we introduce a catalyst-free approach for methane pyrolysis, utilizing direct Joule heating of a porous deposition substrate. This technique involves passing current through a resistive porous carbon cloth to generate heat, effectively breaking down hydrocarbon bonds while producing clean hydrogen and graphitic carbon. We note that the deposition substrate itself is the Joule-heated element, in contrast to indirect Joule heating of a containing device²⁷ used for post-processing methane-pyrolyzed carbon black particles. As methane passes through the carbon cloth, carbon accumulates as concentric multi-layer graphite around the original carbon fiber. We report thorough parametric variations to investigate the impact of power (temperature), pressure, methane flow rate, and reaction duration on hydrogen and graphite production. Through the use of in situ mass spectrometry (MS) and laser absorption spectroscopy (LAS) as monitoring and feedback tools, we demonstrate methane conversion rates up to 88%. Our findings reveal uniform high-quality graphite deposition, achieving a I_D/I_G ratio of 0.1 and 3.38 Å d-spacing within the graphite. The reported process conditions provide a first baseline for this graphite synthesis method that should form the basis for future process and material optimization.

Methodology

Experimental setup

A custom Joule heating fixture integrated into a cold-wall reactor with auxiliary components forms the experimental system and is employed to demonstrate the synthesis of graphitic carbon and hydrogen (see Figure 1). The experimental system comprises a stainless-steel reactor with a 25.4 cm quartz viewport, which was also utilized in previous studies.^29,30,31,32 Other auxiliary components in the system include gas supply mass flow controllers (MKS, GM50A) calibrated for regulating the CH₄ flow rates up to 2,000 sccm. Components for achieving the pressure condition in the chamber include a capacitance manometer (MKS, 624F) measuring up to 1,000 torr, a rotary vane vacuum pump (Oerlikon-Leybold, D65BCS), and an integrated controller (MKS, 946) for process pressure control via an exhaust throttle valve (MKS, T3BI).

Figure 1.

Overview of the Joule heating experimental system for hydrogen and graphitic carbon production from thermal methane decomposition

The system consists of a water-cooled reactor containing a Joule heating fixture with several instruments for temperature and pressure measurement, electrical power, PID control, vacuum control, reactant gas supply, residual gas analysis, and laser absorption spectroscopy.

Joule heating is achieved by applying electrical current to a conductive carbon cloth substrate. Two DC power supplies operating in parallel at 30 V and totaling 200 A (XP Power, HDS3000PS30) are utilized for heating. To maintain consistent power during the experiment, signals from voltage (Aim Dynamics, AIMDC-10V-100V) and current (Phoenix Contact, MCR-SL-CUC-200-U) transducers are acquired. A proportional-integral-derivative (PID) controller (NI, USB-6351) is utilized to adjust the power supply in response to changes in substrate resistance caused by deposition. The control logic adheres to specified constraints as follows:

$$P=\text{Const}.=VI=I^{2}R$$ (Equation 2)

$$\text{with}\text{carbon}\text{deposition}R\downarrow =V\downarrow /I\uparrow$$ (Equation 3)

The Joule heating fixture (see Figure 2) is fabricated using high-temperature tungsten electrode materials: rods with dimensions of 12.7 mm diameter and 152.4 mm length, as well as sheets measuring 3.175 mm in thickness (Midwest Tungsten Service, 40300-PureWRod and 42003-WSheet). Between these components, the carbon cloth is positioned to serve as the current pathway and functions as a resistive heater. The electrodes are situated on alumina (Al₂O₃) sheets, providing both structural support and electrical/thermal insulation within the assembly (Zircar Refractory Composites, Inc, RS-99R).

Figure 2.

Schematic of the Joule heating experimental fixture design with a detailed summary of components

To minimize heat losses from the substrate, radiation shields are fabricated from 0.762-mm-thick molybdenum sheets (Midwest Tungsten Service, 42003-molysheet), positioned to face the upper and lower carbon cloth surfaces. Zirconia (ZrO₂) parts (Zircar Refractory Composites, Inc, NS01-A) are employed to seal the edges of the carbon cloth. Additionally, zirconia provides electrical insulation between the electrodes and mitigates heat loss from the sample. An illustrative overview of the Joule heating fixture is depicted in Figure 2.

Materials characterization

The carbon fabric (FuelCellEarth, CCP100) used in this study is woven from spun yarn carbon fabrics. It measures 0.38 mm in thickness with areal weight of 0.12 kg/m², has a bulk density of 1,750 kg/m³,³³ and comprises polyacrylonitrile carbon fibers with diameters of 8.9 ± 0.7 μm. According to the vendor, the fabric undergoes a vacuum graphitization process and extended high-temperature soaking, resulting in a thermally stable and chemically consistent product with a low oxidation rate. The cloth was used as-received, without any additional treatment, consistent with prior work.³⁴ Vendor specifications report a carbon content of at least 99%. A ZEISS Supra 40VP field emission scanning electron microscope, equipped with both secondary electron and backscattered electron detectors, was employed to study the morphology of the carbon substrate. Nova Nano 230 scanning electron microscope, equipped with an attached Noran 7 energy dispersive X-ray spectroscopy (EDS) system from Thermo Fisher, was used for surface-level elemental analysis. Raman spectra were obtained using RENISHAW inVia confocal Raman microscope. This microscope operates with a 488-nm laser, a 50× magnification objective lens, and a Centrus CCD detector. The carbon product’s quality and structure are typically assessed from its Raman spectrum, particularly by examining the intensity ratio of the D peak (near 1,350 cm⁻¹) to the G peak (near 1,580 cm⁻¹), as well as the presence or absence of a 2D peak (near 2,700 cm⁻¹).³⁵ A low D/G peak ratio and the presence of a clear, narrow 2D peak suggest the formation of high-quality graphitic material. The in-plane crystallite size L_a is related to the I_D/I_G ratio, as described by the Tuinstra-Koenig relation I_D/I_G = c(λ_ex)/L_a.³⁶ This relation indicates that L_a shows an increasing trend under graphitization.

A similar evaluation can be made using the X-ray diffraction (XRD) spectrum, where a more defined (002) peak and a shift toward the peak position of crystalline graphite at 2θ = 26.5° indicate the production of high-quality graphitic material.³⁷ XRD analysis utilized a Panalytical X’Pert Pro X-ray powder diffractometer with a Cu Kα source (λ_XRD = 1.54 Å). The STAR Methods section provides additional details of scanning electron microscopy (SEM), EDS, and XRD methods.

Gas monitoring

Laser absorption spectroscopy

Gas-phase products in this study were characterized via LAS. The theory of LAS has been extensively outlined in prior work,³⁸ and only a brief discussion is included here to clarify the nomenclature and important results. Specific species mole fractions, X_abs, can be related to the intensity of transmitted monochromatic light by the Beer-Lambert law as

$${\alpha }_v=-\ln {\left(\frac{I_t}{I_0}\right)}_{\nu }=\sum _{i}P{x}_{\text{abs}}{S}_i\left(T\right){\phi }_i\left(\nu ,T,P,X_{\text{abs}}\right)L$$ (Equation 4)

Here, α_ν is the spectral absorbance, I_t/I₀ is the ratio of the experimentally measured transmitted and incident laser intensity at frequency ν (cm⁻¹), P (atm) is the total pressure of the measured gas, S_i(T) (cm⁻²/atm) is the temperature-dependent linestrength of transition i at temperature T (K), L (cm) is the pathlength of the absorbing gas, and φ_i (cm) is the spectral lineshape of transition i. Integration over the spectral domain of a well-isolated transition yields the absorbance area, A_i

$$A_i={\int }_{-\infty }^{\infty }{\alpha }_{v}d\nu =P{S}_i\left(T\right)x_{\text{abs}}L$$ (Equation 5)

Through this integration, the lineshape dependence of the absorbance area is removed. In practice, this integration is achieved by fitting a Voigt lineshape to measured spectra. Finally, with independent knowledge of the pressure, temperature, and pathlength of the absorbing gas, the species-specific mole fraction may be obtained.

In this work, three mid-infrared-distributed feedback interband cascade lasers were leveraged to probe the strongly absorbing fundamental C-H stretch vibrational bands intrinsic to hydrocarbons. Specifically, rovibrational transitions in the 3.00–3.34 μm range were selected for the four most abundant hydrocarbons during the decomposition of CH₄. Quantification of these hydrocarbons (CH₄, C₂H₂, C₂H₄, and C₂H₆) facilitates indirect measurement of H₂ through a molar balance. Details on this method and the spectroscopic techniques employed to perform these measurements are extensively available in prior work³⁹ and omitted here in the interest of brevity. The relative uncertainty in the mole fraction measurements of each species is 2.5% for CH₄, 4% for C₂H₂, 11% for C₂H₄, and 3% for C₂H₆ at the typical experimental condition. The uncertainty in the inferred mole fraction of H₂ is estimated to be approximately 4%, obtained by summing the measurement uncertainties of each hydrocarbon in quadrature.³⁹

Mass spectroscopy

A compact, high-resolution residual gas analyzer (RGA) (INFICON, TSPTT200), functioning as an in situ mass spectrometer, was set up to quantify the composition of the product stream in addition to the LAS sensor.⁹ The mass spectrometer was calibrated to quantify relative mole fractions of key product species formed during methane decomposition, including H₂, CH₄, C₂H₂, C₂H₄, and C₂H₆. The calibration gas stream compositions were controlled using mass flow controllers under conditions resembling experimental settings, considering potential variations in species diffusion through the leak valve and into the RGA compartment. MS calibration details are provided in the STAR Methods section. Upon calibration, uncertainty estimates for different species based on relative residuals were determined as follows: H₂ = ± 1.1%, CH₄ = ± 2.9%, C₂H₂ = ± 5.5%, C₂H₄ = ± 24%, and C₂H₆ = ± 34%. The larger uncertainties in minor species in the product stream, C₂H₂, C₂H₄, and C₂H₆, are due to their dilute concentrations, resulting in low signal-to-noise ratios. The LAS sensor provides more reliable measurements of the minor hydrocarbons, while the MS system directly detects H₂.

Temperature measurement

The porous carbon substrate temperature is estimated using an emission spectroscopy technique.⁴⁰ A Horiba iHR 550 imaging spectrometer with a 1,200 g/mm blazed holographic grating, CCD camera (Synapse Plus OE, Horiba)/InGaAs arrays (Symphony II Linear IGA, Horiba), and compact CCD spectrometers (CCS175, Thorlabs) are used to acquire the spectral radiation intensity from the carbon substrate. The emitted light is collected through a calcium fluoride (CaF₂) viewport (KJL, VPZL-275UC) from surface emittance and directed into the spectrometer slit, utilizing a fiber-optic cable (Thorlabs, FG400AEA) and focusing lenses. To ensure the accuracy of temperature measurements, the spectrometer and grating are calibrated using a laser with a single wavelength centered at 532 nm.

Given that accurate temperature measurements rely on radiation intensity at specific wavelengths, the emission intensity depicted in Figure 3 is fitted to Planck’s distribution⁴¹ based on the gray body assumption of the sample in the wavelength range of temperature measurement.^42,43 This fit utilizes the measured intensity, I, as a function of wavelength, λ, through the following⁴⁴:

$$E_{\lambda ,b}\left(\lambda ,T\right)=\frac{C_1}{{\lambda }^5\left[\exp \left(C_2/\lambda T\right)-1\right]}$$ (Equation 6)

$$I\left(\lambda ,T\right)=Aϵ\left(\lambda \right)E_{\lambda ,b}\left(\lambda ,T\right)=B{E}_{\lambda ,b}\left(\lambda ,T\right)$$ (Equation 7)

where $C_1=2\pi h{c}_o^2=3.742\times {10}^8\mathrm{W}·\mu {\mathrm{m}}^4/{\mathrm{m}}^2$, C₂ = hc_o/k_B = 1.439 × 10⁴ μm K, ε represents the emissivity of the carbon substrate, h denotes Planck’s constant, c₀ represents the speed of light (ms⁻¹) in vacuum, and k_B is the Stefan-Boltzmann constant. Temperature and the constant parameter B are determined through the fitting procedure.

Figure 3.

The emission spectrum from a heated carbon cloth for temperature measurements

Chemical conversion and yield

Methane decomposition involves an endothermic reaction that breaks down methane into hydrogen gas and solid carbon in the global reaction²⁶

$${\text{CH}}_4\to \mathrm{C}+2{\mathrm{H}}_2\left(\Delta H^o=74.6\text{kJ}/\text{mol}\right)$$ (Equation 8)

where ΔH is the standard molar enthalpy of the reaction. This global reaction involves a stepwise dehydrogenation mechanism that includes light hydrocarbons as intermediaries¹⁰

$$2{\text{CH}}_4\stackrel{-{\mathrm{H}}_2}{\to }{\mathrm{C}}_2{\mathrm{H}}_6\stackrel{-{\mathrm{H}}_2}{\to }{\mathrm{C}}_2{\mathrm{H}}_4\stackrel{-{\mathrm{H}}_2}{\to }{\mathrm{C}}_2{\mathrm{H}}_2\stackrel{-{\mathrm{H}}_2}{\to }2\mathrm{C}$$ (Equation 9)

Measurement of the product stream composition by both MS and LAS techniques facilitate determination of the hydrogen yield ($Y_{{\mathrm{H}}_2}$), carbon yield (Y_C), and methane conversion ($X_{{\text{CH}}_4}$) of the process. These metrics are calculated using the mole fractions of the five most prominent species in the product stream, each representing 0.1% or more of the total composition, which includes H₂, CH₄, C₂H₂, C₂H₄, and C₂H₆. Importantly, this approach enables a time-resolved quantification of these metrics without implementation of a steady-state assumption as in prior similar work.⁴⁵ By establishing a balance over the hydrogen atoms between the feedstock and product streams, the exhaust molar flow rate is established as

$${\dot{n}}_{\text{out}}=\frac{2{\dot{n}}_{{\text{CH}}_4,\text{in}}}{x_{{\mathrm{H}}_2}+2x_{{\text{CH}}_4}+x_{{\mathrm{C}}_2{\mathrm{H}}_2}+2x_{{\mathrm{C}}_2{\mathrm{H}}_4}+3x_{{\mathrm{C}}_2{\mathrm{H}}_6}}$$ (Equation 10)

where the inlet molar flow rate of CH₄ is calculated as ${\dot{n}}_{{\text{CH}}_4,\text{in}}={\dot{V}}_{{\text{CH}}_4}{\overline{\rho }}_{{\text{CH}}_4}$. ${\dot{V}}_{{\text{CH}}_4}$ (sccm) is the volumetric flow rate regulated by the mass flow controllers, and ${\overline{\rho }}_{{\text{CH}}_4}$ (mol cm⁻³) is the standard density of methane gas. Because the hydrogen mole fraction, $x_{{\mathrm{H}}_2}$, is not measured directly using the LAS sensor, it is indirectly calculated by employing the major products assumption as

$$x_{{\mathrm{H}}_2}=1-\left(x_{{\text{CH}}_4}+x_{{\mathrm{C}}_2{\mathrm{H}}_2}+x_{{\mathrm{C}}_2{\mathrm{H}}_4}+x_{{\mathrm{C}}_2{\mathrm{H}}_6}\right)$$ (Equation 11)

After acquiring the outlet molar flow rate, methane conversion quantifies the relative amount of consumed CH₄ as

$$X_{{\text{CH}}_4}=\frac{{\dot{n}}_{{\text{CH}}_4,\text{in}}-{\dot{n}}_{\text{out}}{x}_{{\text{CH}}_4}}{{\dot{n}}_{{\text{CH}}_4,\text{in}}}$$ (Equation 12)

Likewise, total hydrogen yield, representing the proportion of inlet methane that appears as hydrogen gas in the product stream, is determined from

$$Y_{{\mathrm{H}}_2}=\frac{{\dot{n}}_{\text{out}}{x}_{{\mathrm{H}}_2}}{2{\dot{n}}_{{\text{CH}}_4,\text{in}}}$$ (Equation 13)

Similar to the overall hydrogen yield, the total carbon yield is

$$Y_{\mathrm{C}}=\frac{{\dot{m}}_{\mathrm{C}}}{M_{\mathrm{C}}{\dot{n}}_{{\text{CH}}_4,\text{in}}}$$ (Equation 14)

where the rate of carbon mass deposition, ${\dot{m}}_{\mathrm{C}}$ (kgs⁻¹), derived from a balance of carbon atoms, is represented as

$${\dot{m}}_{\mathrm{C}}=\left[{\dot{n}}_{{\text{CH}}_4,\text{in}}-{\dot{n}}_{\text{out}}\left(x_{{\text{CH}}_4}+2x_{{\mathrm{C}}_2{\mathrm{H}}_2}+2x_{{\mathrm{C}}_2{\mathrm{H}}_4}+2x_{{\mathrm{C}}_2{\mathrm{H}}_6}\right)\right]M_{\mathrm{C}}$$ (Equation 15)

where M_C (kg/kmol) is the molecular weight of a carbon atom.

Efficiency quantification and residence time

The electrical-to-chemical (ETC) efficiency in methane pyrolysis is calculated under the assumption of ideal direct decomposition of methane without side reactions or by-products (specifically, CH₄ converts only to H₂ and C), while also ignoring the sensible heating of unconverted methane^46,47 as

$${\eta }_{\text{ETC}}=\frac{X_{{\text{CH}}_4}{\dot{n}}_{{\text{CH}}_4,\text{in}}\left[{\int }_{T_{\text{in}}}^{T_{\mathrm{R}}}{\overline{c}}_{p,\text{CH}4}\left(T\right)\mathrm{d}T+\Delta H_{\mathrm{R}}\left(T_{\mathrm{R}}\right)\right]}{{\dot{Q}}_{\text{Joule}}}$$ (Equation 16)

This metric defines ΔH_R as the molar enthalpy change at the average reaction temperature T_R, where ${\overline{c}}_p$ represents the molar heat capacity and $X_{{\text{CH}}_4}$ denotes the extent of methane conversion. Temperature-dependent properties of fluid and solid carbon were sourced from existing literature to calculate the molar enthalpy change of the reaction.^48,49,50

The estimation of the mean residence time of the flow involves the assumption that the flow consists entirely of methane traversing the entire reaction zone within the porous carbon medium. By utilizing the porosity of the fibrous medium (ψ), the mean residence time (t_res) is calculated through

$$t_{res}=\frac{{\rho }_{{\text{CH}}_4}\left(T_{\mathrm{R}}\right)V_{\mathrm{R}}\psi }{\dot{m}}$$ (Equation 17)

where ψ = 0.827 and V_R represents the volume of the reaction zone, calculated as V_R = LWδ, where L, W, and δ denote the length, width, and thickness of the reaction zone, respectively. The symbol $\dot{m}$ indicates the mass flow rate of methane at the inlet, while ${\rho }_{{\text{CH}}_4}\left(T_{\mathrm{R}}\right)$ signifies the density of CH₄ gas at the measured reaction temperature.

Results and discussion

In this section, results from a parametric investigation of this direct heating pyrolysis process are detailed. Specifically, the effect of variations in heating input power, chamber pressure, feedstock flow rate, and experimental duration on product yields and quality were investigated. In each experiment, the reactor is initially evacuated to approximately 10 mTorr to ensure a leak-free environment. Subsequently, CH₄ gas (99.999% purity) is introduced at the prescribed flow rate, and the reactor is allowed to achieve the desired pressure setpoint prior to initiation of substrate heating.

The original carbon cloth exhibits characteristics typical of low-quality carbon with some local graphitic domains, producing a high D/G peak ratio (I_D/I_G = 0.89) and a wide 2D peak. Similar observations are also apparent in the XRD spectra of the initial carbon cloth with no distinct narrow XRD peaks. The (002) reflection position (2θ₀₀₂) of the original carbon medium is 26.2°, with a full width at half maximum (FWHM) of 3.2°. Applying Bragg’s law and the Scherrer equation³⁷ detailed in the STAR Methods section, the average interplanar distance and average crystallite size along the c axis of the original product measure 3.40 Å and 2.56 nm, respectively. The initial fiber diameter is approximately 8.9 ± 0.7 μm.

The porous carbon cloth’s thin profile, symmetrical design of the fixture and radiation shields, and uniform substrate heating between the two electrodes ensure consistent spatial process temperature and chemical reactions. The process in this work rapidly stabilizes under steady-state conditions, often within few minutes of heating from room temperature (refer to Figure 4A for visuals).

Figure 4.

Electrical, thermal, and chemical parameters variations with time

(A) Data collected from the Joule heating experiment for methane flow rate of 100 sccm, pressure of 25 torr, and Joule heating power of 2.5 kW including substrate temperature, power input, resistivity, input voltage, and current.

(B) Data include the mole fraction of major species, methane conversion ($X_{{\text{CH}}_4}$), and hydrogen ($Y_{{\mathrm{H}}_2}$) and carbon (Y_C) yields for both MS and LAS (MS data: o, LAS data: x).

Some studies have employed direct solar heating for synthesizing high-quality graphene^30,31 and producing graphitic carbon and hydrogen^9,26 within a short time frame. In contrast, most previous studies on methane pyrolysis using indirect heat sources such as a hot furnace⁵¹ or solar absorber²² reported achieving steady-state thermal conditions after approximately 30 min or more. Here, we investigate direct Joule heating of the substrate, which produces uniform carbon deposition, rapid processing conversions, short run times, and high temperatures constrained only by the power supply and material integrity.

The nominal methane pyrolysis process conditions are methane flow rate of 100 sccm, pressure at 25 torr, and Joule heating power of 2.5 kW. Figure 4A depicts the variation of substrate temperature over time, while the power remains constant through the PID controller operation. The process shows changes in resistivity caused by graphitic carbon deposition during methane pyrolysis. With increasing deposition of conductive graphitic carbon material, the resistivity decreases, resulting in a drop in voltage and an increase in current to maintain consistent power delivery, as illustrated in Figure 4A.

Inlet methane rapidly converts into primarily hydrogen in the product stream, while high-quality graphitic carbon accumulates concentrically around the fibers of the fibrous porous carbon substrate. Transient mole fractions in the product stream are presented in Figure 4B by MS and LAS techniques. Within 10 min, the hydrogen mole fraction reaches its maximum value ($x_{{\mathrm{H}}_2}$ = 87%), alongside unconverted methane ($x_{{\text{CH}}_4}$ = 7.9%) analyzed by MS. A secondary by-product, acetylene ($x_{{\mathrm{C}}_2{\mathrm{H}}_2}$ = 4.0%), also forms. Lower concentrations of ethylene and ethane with a combined mole fraction ($x_{{\mathrm{C}}_2{\mathrm{H}}_4}$ and $x_{{\mathrm{C}}_2{\mathrm{H}}_6}$) of 1.1% are also produced. The LAS results indicate that the hydrogen mole fraction ($x_{{\mathrm{H}}_2}$) is 88.5%, with a residual methane mole fraction ($x_{{\text{CH}}_4}$) of 7.9%. Additionally, acetylene ($x_{{\mathrm{C}}_2{\mathrm{H}}_2}$) is produced with a mole fraction of 3.3%. Lower concentrations of ethylene and ethane are also present, with a combined mole fraction ($x_{{\mathrm{C}}_2{\mathrm{H}}_4}$ and $x_{{\mathrm{C}}_2{\mathrm{H}}_6}$) of 0.3%.

The significant presence of C₂H₂ in the product stream decreases the carbon yield from methane conversion more than the hydrogen yield, as C₂H₂ has a lower H:C ratio compared to CH₄. Thus, for the conditions shown in Figure 4B, MS analysis shows $X_{{\text{CH}}_4}$ = 86%, $Y_{{\mathrm{H}}_2}$ = 80%, and Y_C = 67%. On the other hand, LAS reveals $X_{{\text{CH}}_4}$, $Y_{{\mathrm{H}}_2}$, and Y_C of 85%, 82%, and 72%, respectively. In addition to the uncertainties in MS and LAS techniques, Fincke et al.¹⁹ reported that methane equilibrium pyrolysis around 2,000°C results in the presence of hydrogen and ethynyl radicals in the product stream. The discrepancies between MS and LAS data in this temperature range for methane pyrolysis arise from two factors: the LAS technique estimating the hydrogen mole fraction as the remainder after accounting for other gases and the larger uncertainties in the MS technique for minor species, specifically C₂H₄ and C₂H₆, due to their low concentrations and resulting low signal-to-noise ratios. While the LAS sensor provides more reliable measurements of these trace species, the MS system directly measures the hydrogen mole fraction but overestimates the trace species concentrations, leading to minor discrepancies in the results. The potential error sources and uncertainties in both gas composition measurement techniques (MS and LAS) have been detailed in prior work.^9,39 The discrepancies in the two measurement techniques, particularly apparent at the early stages of decomposition, likely result from the transit time of the exhaust gas to the two separate measurement locations. However, as the process approaches steady state, and the exhaust gas composition evolves more gradually, the discrepancy due to differing transit times is minimized and both measurement techniques show good agreement.

The following sections describe the effects of heating power, pressure, and flow rate on graphitic carbon and hydrogen production. For the effect of duration, refer to the Supplemental Information (Figures S1–S3).

Effect of heating power

The effect of heating power on the production of graphitic carbon and hydrogen during methane pyrolysis is analyzed here. The heating power was adjusted to 1.5, 2.0, 2.5, and 3.0 kW, resulting in measured temperatures of 1,825, 1,874, 2,145, and 2,358 K, respectively. All other conditions were held constant, with the pressure at 25 torr, the methane flow rate at 100 sccm, and the duration at 10 min throughout the experiment. The maximum power level reached in this study was 3.0 kW. However, due to the rapid decomposition of methane, the power supply current saturated at 210 A, reaching the power supply limit. Beyond 7 min, it became infeasible to sustain the power at 3.0 kW.

Increased power levels result in broader deposition areas compared to lower power settings, where a narrower strip forms between the electrodes. Lower power levels lead to carbon being initially deposited along the centerline, mainly because this region experiences higher temperatures due to symmetrical heat transfer profile from insulated edges, where the current naturally follows the path of least resistance. At higher power levels, the abundance of power supply and elevated temperatures promote a wider spread of carbon deposition across the larger width of the carbon cloth.

The solid carbon produced through the Joule heating pyrolysis process mainly comprises high-quality graphitic material. Cross-sectional SEM imaging reveals the morphology and the impact of increasing power on graphite deposition around the original fiber, as shown in Figures 5A–5D. As previously mentioned, power affects the deposition area. Although there is a decreasing trend in the deposition diameter with increasing power, measuring 48, 44, 28, and 13 μm respectively, the total deposited weight on the carbon cloth amounts to 0.141, 0.259, 0.373, and 0.359 g over the entire carbon cloth.

Figure 5.

Effect of power variation on graphite product

(A–D) SEM images depict the overall morphology of deposited graphite and provide a cross-sectional fiber view at different power levels (1.5, 2, 2.5, and 3 kW), corresponding to measured temperatures of 1,825, 1,874, 2,145, and 2,358 K, respectively. (E) Raman spectra, highlighting the I_D/I_G ratio. (F) XRD spectra of the product powder, indicating inter-layer spacing and crystalline size. All parameters other than power were held constant at 25 torr, a CH₄ flow rate of 100 sccm, and a duration of 10 min (7 min for 3 kW). Scale bars: 20 μm (overall morphology) and 2 μm cross-sectional view.

Graphene forms the fundamental structure for depositing multi-layer graphite. According to Kim et al.⁵² the creation of graphite layers arises from the crystallization of a supersaturated carbon-adatom species. The density of nucleation is affected by temperature-dependent processes, with activation energies ranging from 1 to 3 eV. As temperature increases, the decrease in nucleation density (resulting in larger grain sizes) is associated with a higher probability of capturing supercritical carbon nuclei from the gas feed rather than initiating nucleation on recently available substrate sites due to heightened desorption rates.⁵²

As the power level increases from 1.5 to 2 and 2.5 kW, the temperatures reach 1,825, 1,874, and 2,145, respectively. Consequently, the size of graphite grains also increases, resulting in declining Raman I_D/I_G ratios of 0.18, 0.14, and 0.10 and increased crystalline size measured by XRD of 7.10, 9.45, and 10.3 nm, respectively, as depicted in Figures 5E and 5F. However, at a power level of 3 kW (corresponding to 2,358 K), substrate resistivity decreases further due to a wider graphite deposition area (43 × 44 mm), with the process time constrained to 7 min by the power supply, resulting in a lower ratio of deposited graphite compared to the original carbon fiber per unit area and a smaller deposition diameter (12.6 μm). Consequently, the I_D/I_G ratio increases slightly to 0.15, where laser heating during Raman spectroscopy^53,54 of graphite can cause localized temperature increases, particularly in thinner layers, which are less effective at dissipating heat. This can lead to defects and alter the crystalline structure. XRD analysis of the powderized sample reveals an intermediate crystalline size of 8.11 nm, likely associated with the lower-quality original carbon cloth. Despite carbon cloth quality improvements from high-temperature annealing, this lower-quality material may still contribute to the overall reduced crystal size.

As the power level increases from 1.5 to 3 kW, the deposition weight, deposition area, and carbon yield increase from 0.141 g, 44 × 16 mm, and 34% to 0.359 g, 43 × 44 mm, and 72%, respectively. This occurs despite a decrease in fiber diameter size from 48 to 13 μm, as shown in Table S1 and Figure 6B. Nevertheless, at 3 kW (corresponding to 2,358 K), graphite displays an average interplanar distance d₀₀₂ of 0.338 nm, consistent with previous studies on graphite processing above 2,000°C, whereas d₀₀₂ in processes below 2,000°C falls within the range of 0.340–0.341 nm.⁵⁵

Figure 6.

Effect of power variation on species concentrations, methane conversion, and hydrogen/carbon yields

Gas analysis from RGA mass spectrometry for methane decomposed at different power levels (1.5, 2 , 2.5, and 3 kW) showing (A) major species concentration (χ) and (B) methane conversion ($X_{{\text{CH}}_4}$), hydrogen yield ($Y_{{\mathrm{H}}_2}$), and carbon yield (Y_C). All other parameters were held constant at 25 torr, a CH₄ flow rate of 100 sccm, and a duration of 10 min (7 min for 3 kW).

The process conversion and yields notably improve with increased heating power from 1.5 to 3.0 kW, resulting in a temperature rise from 1,825 to 2,358 K. Figures 6A and S4A depict the mole fractions using MS and LAS methods, respectively. The data in Figures 6B and S4B show conversions and yields for the MS and LAS methods, respectively. The MS technique reveals a substantial increase in $X_{{\text{CH}}_4}$ from 61% to 88%, $Y_{{\mathrm{H}}_2}$ from 53% to 82%, and Y_C from 34% to 72% as power increases as shown in Figure 6B.

Heating power plays a crucial role in methane pyrolysis, influencing temperature and subsequent conversion rates. The process temperatures observed are slightly higher than those reported in prior literature.^22,26 However, residence times are drastically lower by an order of magnitude for similar methane conversions in prior work,^22,26 with values increasing from 0.71 ms at 1.5 kW to 1.5 ms at 3.0 kW due to increased deposition area with power. This indicates swift processing with high conversion and yield rates, possibly due to the improved direct Joule heating of the porous medium volume.

For process performance, the electrical-to-chemical efficiency (η_ETC) starts at 0.61% at 1.5 kW and decreases to 0.55% at 3.0 kW. This trend is attributed to the phenomenon that increasing power leads to greater thermal losses to the surroundings, resulting in reduced efficiency. Although higher process conversion and yields are achieved at higher power levels, this increase does not translate to improved process efficiency.

Effect of pressure

The influence of pressure on the yield of graphitic carbon and hydrogen during methane pyrolysis was investigated. Pressures were adjusted to 10, 25, 50, and 100 torr, with all other conditions remaining fixed: power was set at 2.5 kW, the methane flow rate at 100 sccm, and the duration maintained at 10 min throughout the experiment. Higher pressures lead to narrower deposition areas (43 × 15 mm) in comparison to lower-pressure settings, where a broader strip (43 × 22 mm) forms between the electrodes. The increased pressure causes carbon to initially deposit along the centerline, primarily because this region experiences higher temperatures due to symmetrical heat transfer from the edges, where the current naturally follows the path of least resistance. For lower-pressure conditions, the lower concentration of methane and higher temperatures promote a wider spread of graphite across the larger width of the carbon cloth.

Cross-sectional SEM imaging provides insights into the morphology and illustrates the impact of increasing pressure on the deposition of the graphite annulus on the original fiber, as depicted in Figures 7A–7D. As noted previously, pressure plays a role in determining the deposition area. An increasing trend in the deposition diameter size occurs with pressure measuring 14, 28, 82, and 140 μm, respectively, with corresponding deposited weights on the carbon cloth of 0.284, 0.373, 0.471, and 0.537 g.

Figure 7.

Effect of pressure variation on graphite product

(A–D) SEM images depict the overall morphology of deposited graphite and provide a cross-sectional fiber view at different pressure levels (10, 25, 50, and 100 torr), corresponding to measured temperatures of 2,262, 2,145, 2,473, and 2,280 K, respectively. (E) The Raman spectra, highlighting the I_D/I_G ratio. (F) The XRD spectra of the product powder, indicating inter-layer spacing and crystalline size; all parameters other than pressure were held constant at 2.5 kW, a CH₄ flow rate of 100 sccm, and a duration of 10 min. Scale bars: 20 μm (overall morphology) and 2 μm (cross-sectional view).

The synthesis of graphitic layers under low-pressure conditions appears to be governed by surface reactions that are sensitive to temperature uniformity. Under these circumstances, the flux of active species decreases, leading to a reduction in collisions and an increase in diffusion.⁵⁶ Conversely, higher pressure imposes limitations on the diffusion of active species through the boundary layer (BL). Consequently, the local gas flow patterns and reactor geometry have a significant impact on the process, resulting in nonuniform growth of graphitic layers.⁵⁶

As the pressure increases to 10, 25, 50, and 100 torr, the size of graphite grains also increases up to a certain limit, beyond which the quality begins to deteriorate. This effect is attributed to the pressure’s influence on the flow distribution and substrate temperature, which affect the associated deposition process. The changes in grain size are evident in the Raman spectra, showing a reduced I_D/I_G ratio of 0.25 and 0.10 initially, followed by an increase to 0.12 and 0.32, respectively. Similarly, the characteristic crystal size estimated by XRD initially trends upward to 8.44 and 10.3 nm before declining to 9.58 and 7.61 nm, as illustrated in Figures 7E and 7F.

The effect of pressure on graphite deposition is also influenced by the convective cooling effect and the area of graphite deposition acting as a resistive path for Joule heating. Specifically, at 10 and 25 torr, similar deposition areas are observed with a deposition width of 22 mm, resulting in decreased temperatures of 2,262 and 2,145 K, respectively, due to increased cooling with pressure. On the other hand, at 50 and 100 torr, similar deposition areas are observed with a deposition width in the range of 13–15 mm, with temperatures showing a decreasing trend to 2,473 K and 2,280 K, respectively, for the same reasons mentioned above.

Furthermore, increasing pressure for the same heated area from 10 to 25 torr leads to an increase in the average interplanar distance d₀₀₂ to 0.339 nm and 0.341 nm, respectively. Similarly, increasing pressure to 50 and 100 torr also results in a spacing increase from 0.338 to 0.340 nm, respectively. The spacing at lower pressures is significantly influenced by the process temperature, as observed in previous studies,⁵⁵ compared to factors such as fiber size or grain sizes.

Initially, a slight increase in process conversion occurs, likely influenced by the pressure increase from 10 to 25 torr and the resulting longer residence time due to increased concentration that impacts methane decomposition into other minor hydrocarbon species (C₂H₄ and C₂H₆). However, beyond 25 torr, the impact of increased pressure on both conversion and yields shows decreasing trends. This phenomenon can be attributed to two competing factors influencing the overall reaction. First, increased pressure tends to shift the decomposition reaction backward in accordance with Le Chatelier’s principle,²⁶ leading to decreased conversion. Second, it results in an increased mean residence time of the gas phase, potentially enhancing the conversion rate. Consequently, decreases in conversion and yield are generally observed as pressure increases up to 100 torr due to the reaction shifting backward. Another issue associated with high pressure is the observed deposition of thick layers of carbon black on the radiation shield ahead of the carbon cloth.

Figures 8A and S5A depict the evolution of mole fractions using MS and LAS methods, respectively. The data in Figures 8B and S5B show conversions and yields for the MS and LAS methods, respectively. The MS technique reveals $X_{{\text{CH}}_4}$ remaining fairly stable in the range 84%–86% with an increase in pressure between 10 and 25 torr within the same deposition area, followed by a decrease from 84% to 74% between 50 and 100 torr, respectively. Meanwhile, yields follow a decreasing trend, with $Y_{{\mathrm{H}}_2}$ decreasing from 80% to 63% and Y_C from 71% to 34% as the pressure increases.

Figure 8.

Effect of pressure variation on species concetrations, methane conversion, and hydrogen/carbon yields

Gas analysis from RGA mass spectrometry for methane decomposed at different pressures (10, 25, 50, and 100 torr) showing (A) major species concentration (χ) and (B) methane conversion ($X_{{\text{CH}}_4}$), hydrogen yield ($Y_{{\mathrm{H}}_2}$), and carbon yield (Y_C). All other parameters were held constant at 2.5 kW, a CH₄ flow rate of 100 sccm, and a duration of 10 min.

Pressure is a key factor in methane pyrolysis, influencing the cooling rate, residence time, reaction equilibrium, and deposition area, consequently affecting the conversion rates. However, residence times show a marked decrease compared to previous studies^22,26 for comparable methane conversions due to the small substrate thickness and direct heating method, with values increasing from 0.31 ms at 10 torr to 2.0 ms at 100 torr. This characteristic suggests fast kinetics with high conversion and yield, likely due to enhanced direct Joule heating of the porous medium volume as discussed earlier. In terms of process performance, the electrical-to-chemical efficiency (η_ETC) starts at 0.61% at 10 torr and slightly increases to 0.65% at 50 torr, attributed to the higher temperature achieved, the deposition pattern, and the resulting increased conversion. However, process efficiency generally exhibits a decreasing trend, reaching 0.54% at 100 torr, mirroring the trend in methane conversion.

Effect of flow rate

The impact of flow rate on the formation of graphitic carbon and hydrogen during methane pyrolysis was assessed. Residence time primarily depends on the mass flow rate, following the relationship $t_{res}\propto 1/\dot{m}$. Four distinct flow rates were tested: 100, 200, 500, and 1,000 sccm, while all other variables were kept constant with power at 2.5 kW, pressure at 10 torr, and a duration of 10 min. The decision to keep the pressure at 10 torr (low pressure) derived from observations showing that graphite tended to deposit over a broader area rather than on a narrower strip at higher pressures. This approach aimed to enhance areal graphite deposition on the carbon cloth by increasing the flow rate or extending the duration, thereby collecting more carbon. This strategy also circumvented the limitation of deposition on a narrower strip, which could saturate the power supply current and hinder a thorough parametric study on flow rate.

Similar to pressure, higher flow rates produce narrower deposition areas (13–15 mm in width) compared to lower-flow-rate settings, where a broader strip (22 mm in width) forms between the electrodes. Higher flow rates promote increased cooling rates and shorter residence times. The increased flow rate causes carbon to be initially deposited along the centerline, primarily because this region experiences higher temperatures due to the symmetrical heat transfer profile from insulated edges, where the current naturally follows the path of least resistance. At lower flow rates, the longer residence time and higher temperatures over a larger area promote a wider spread of carbon across the larger width of the carbon cloth.

Cross-sectional SEM imaging offers valuable insights into the morphology and visually demonstrates the influence of increasing flow rate on the deposition of graphite, as shown in Figures 9A–9D. An increasing trend in diameter follows with flow rate, measuring 14, 29, 34, and 52 μm, respectively, with corresponding deposited weights on the carbon cloth of 0.284, 0.387, 0.503, and 0.571 g.

Figure 9.

Effect of flow rate variation on graphite product

(A–D) SEM images depict the overall morphology of deposited graphite and provide a cross-sectional fiber view at different flow rates (100, 200, 500, and 1,000 sccm), corresponding to measured temperatures of 2,145, 2,165, 2,427, and 2,298 K, respectively. (E) The Raman spectra, highlighting the I_D/I_G ratio. (F) The XRD spectra of the product powder, indicating inter-layer spacing and crystalline size. All parameters other than flow rate were held constant at 2.5 kW, 10 torr, and a duration of 10 min. Scale bars: 20 μm (overall morphology) and 2 μm (cross-sectional view).

At lower flow rates, methane decomposition exhibits high methane conversion and hydrogen yield of 84% and 80%, respectively. However, these values decrease to 29% and 24% at higher flow rates. This variation in flow rate profoundly impacts both conversions and yields, consequently influencing the H₂ to CH₄ ratio. Hydrogen plays a dual role in the growth of graphitic layers: it aids in methane chemisorption by activating surface-bound carbon and acts as an etching agent that regulates grain size and morphology.⁵⁷ Additionally, hydrogen mitigates the adverse effects of stray oxidizing contaminants present in the gas feed or oxidized substrate.

As the flow rate ranges from 100, 200, 500, to 1,000 sccm, a corresponding increase in the size of graphite grains occurs until a threshold is reached, after which the quality begins deteriorating. This phenomenon is attributed to the flow rate’s impact on the BL and the substrate’s temperature, both of which influence the deposition process. These changes in grain size are evident in the Raman spectra, where the I_D/I_G ratio decreases initially to 0.25, 0.20, and 0.17. Subsequently, the I_D/I_G ratio increases from 0.17 to 0.35 at 1,000 sccm, indicating a decline in quality. However, at 500 and 1,000 sccm (higher flow rates), graphite deposition exhibits distinct narrower strip deposition regimes. Thus, the process undergoes competing effects as the flow rate increases: increased temperature due to current flowing on a narrower resistive strip while also introducing a higher cooling rate. Similarly, the crystalline size, as measured by XRD, increases to 8.44, 9.22, and 10.5 nm before decreasing to 10.0 nm, as depicted in Figures 9E and 9F.

The area of graphite deposition is influenced by the flow rate due to a combination of heat convection’s cooling effect and the resistive path for Joule heating created by the deposited graphite. Specifically, at flow rates of 100 and 200 sccm, the same deposition area leads to reduced temperatures from 2,262 to 2,165 K, respectively, due to enhanced cooling. Conversely, experiments conducted at 500 and 1,000 sccm consistently show narrower deposition areas, resulting in decreasing temperatures of 2,427 and 2,298 K, respectively. These temperatures are slightly higher due to the increased ohmic resistance of the narrower strip under constant power.

Furthermore, as the flow rate increases from 100 to 200 sccm, both resulting in the same deposition area, we observe an increase in the average interplanar distance (d₀₀₂) of graphite from 0.339 to 0.343 nm, respectively. This increase is attributed to the combined effect of enhanced cooling and an increased CH₄/H₂ ratio. Further increasing the flow rate to 500 and 1,000 sccm maintains a consistent spacing in the range of 0.340–0.341 nm.

A high flow rate leads to incomplete dissociation of converted methane, resulting in an increased production of acetylene as a by-product with a high CH₄/H₂ concentration. At lower CH₄/H₂ ratios, the growth rates of graphitic layers tend to decrease due to lower carbon concentrations. Prior work has suggested a Volmer-Weber growth mode,⁵⁸ in which the second layer is believed to form below the first layer, with carbon radicals intercalating between the initial graphitic layer and the substrate, resulting in highly oriented stacking. Conversely, turbostratic graphitic layers typically develop on top of the previous layer at higher CH₄/H₂ ratios and elevated process pressures,⁵⁸ and such formations in multi-layer graphene have also been observed from Joule heating on nickel substrates.⁵⁹

A consistent downward trend in methane conversion, as well as in hydrogen and carbon yields, occurs with increasing flow rate. This trend can be attributed to a significant reduction in residence time from 0.31 to 0.022 ms. Figures 10A and S6A illustrate the mole fractions using MS and LAS methods, respectively. The data in Figures 10B and S6B show conversions and yields for the MS and LAS methods, respectively. The MS technique demonstrates a decrease in $X_{{\text{CH}}_4}$ with an increasing flow rate, ranging from 84% to 29% between 100 and 1,000 sccm. Furthermore, the yields exhibit a pronounced decline, with $Y_{{\mathrm{H}}_2}$ decreasing from 80% to 24% and Y_C from 71% to 12% as the flow rate increases. Nonetheless, these reductions are milder than those seen in other types of volumetric and tubular reactors even with particle seeding, likely due to the porous substrate’s ability to improve heat transfer and enable higher temperatures even with shorter residence times.²⁶ Overall, these results indicate that at the temperatures and flow rates evaluated here, there is generally insufficient time for the reaction to proceed to equilibrium, and thus the flow rate or residence time directly determines reaction progress according to the chemical kinetic rates.

Figure 10.

Effect of flow rate variation on species concentrations, methane conversion, and hydrogen/carbon yields

Gas analysis from RGA mass spectrometry for methane decomposed at different CH₄ flow rates (100, 200, 500, and 1,000 sccm) showing (A) major species concentration (χ) and (B) methane conversion ($X_{{\text{CH}}_4}$), hydrogen yield ($Y_{{\mathrm{H}}_2}$), and carbon yield (Y_C). All other parameters were held constant at 2.5 kW, 10 torr, and a duration of 10 min.

The hydrogen production rate approaches a peak of about 2.4 g/h at a flow rate of 1,000 sccm. This increase in H₂ production rate considerably enhances the electrical-to-chemical efficiency, although it comes with a reduced hydrogen yield. As the flow rate rises from 100 to 1,000 sccm, the electrical-to-chemical efficiency improves from 0.61% to 2.1%. These efficiencies correspond to 3,160 and 1,050 kWh of electrical power per kilogram of H₂, respectively. Importantly, the residence time is an order of magnitude less than reported methods in the literature,^22,26 and the energy consumption metric does not take into account the high-value graphitic product generated by the current methane pyrolysis process.

Carbon product and capture

The solid carbon produced by Joule heating pyrolysis is predominantly a high-quality graphitic material. Detailed analysis of the graphitic carbon was performed under conditions that yield the best graphite quality at 2.5 kW (2,473 K), 50 torr, 100 sccm, and 10 min and those aimed at high hydrogen yield at 2.5 kW (2,298 K), 10 torr, 1,000 sccm, and 10 min.

SEM images for the original carbon cloth in Figures 11A and 11B illustrate the woven fibers and their sizes, respectively. After methane pyrolysis under conditions that yield the best graphite quality, the fiber size increases to tens of micrometers due to graphitic deposition, demonstrating significant growth from the original fiber size as shown in Figures 11C, 11D, and 12A. Representative Raman and XRD spectra for both the initial fibrous carbon material and the resultant graphitic carbon product are presented in Figures 11E and 11F. The Raman spectrum of the initial carbon material shows a high D/G peak ratio (I_D/I_G = 0.89). In contrast, the graphitic carbon product exhibits a much lower D/G peak ratio (I_D/I_G = 0.12) and a distinct, narrow 2D peak, indicative of long-range graphitic structures.³⁵

Figure 11.

Comparision between original carbon cloth and deposited graphite

SEM images compare the original carbon cloth (A) and deposited graphite (C). (B and D) show the corresponding higher-magnification views of the overall morphology and cross-section, respectively. (E) Raman spectra with the I_D/I_G ratio.

(F) XRD spectra of the product powder, indicating inter-layer spacing and crystalline size. Graphite synthesis via methane pyrolysis was conducted at 2.5 kW, 50 torr, CH₄ flow rate of 100 sccm, and a duration of 10 min. Scale bars; 100 μm (overall morpology) and 20 μm (higher-magnification views).

Figure 12.

Visualization and characterization illustrate the spatial and quality uniformity of deposition under conditions that yield the best graphite quality at 2.5 kW (2,473 K), 50 torr, 100 sccm, and 10 min

SEM image (A) shows the cross-section of the fiber composite after deposition, while photograph (B) captures the carbon substrate post-deposition. Raman mapping characterization (C and D) details I_D/I_G, I_2D/I_G, and FWHM of the 2D peak along the x and y directions, indicating uniform and high-quality deposition throughout. Scale bars: 10 μm (cross-sectional view) and 5 mm (photographic image).

Similar findings appear in the XRD spectra of the original carbon medium and the graphitic product. The (002) reflection position (2θ₀₀₂) of the initial carbon medium is 26.2°, with a peak FWHM of 3.2°. The original product has an average interplanar distance of 3.40 Å and an average crystallite size along the c axis of 2.56 nm. In comparison, the graphitic product (with 2θ₀₀₂ and FWHM of 26.3° and 0.85°, respectively) has an interplanar distance of 3.38 Å and an average crystallite size of 9.58 nm, indicating significantly higher crystallinity in the graphitic product. Additionally, energy-dispersive X-ray spectroscopy (EDS) analysis reveals carbon energy peaks at approximately 0.277 keV (Kα), demonstrating that the carbon content is at least 99.7% (see Figure S7B).

Joule heating produces high uniformity in graphite deposition and substrate temperature due to the volumetric heating along the current path. Figure 12B shows a photograph of the graphitic carbon deposition area on the intact final substrate, which appears silverish. Raman mapping of the substrate’s spatial distribution reveals a large deposition area measuring 45 mm in length (between the two electrodes) and 13 mm in width with an I_D/I_G ∼ 0.1 as shown in Figures 12C and 12D, respectively.

The graphitic carbon produced during high hydrogen production exhibits a D/G peak ratio (I_D/I_G = 0.35), an interplanar distance of 3.40 Å, and an average crystallite size of 10.0 nm. The fiber size increases to tens of micrometers, demonstrating significant growth from the original fiber size as shown in Figure S8A. According to the EDS analysis, energy peaks of carbon occur at 0.277 keV (Kα), and the carbon content is at least 99.6% (see Figure S7D). Figure S8B contains a photograph of the graphitic carbon deposition area on the intact final substrate, which appears silverish. Similarly, Raman mapping of the substrate’s spatial distribution reveals a large deposition area measuring 43 mm in length and 16 mm in width with an I_D/I_G ∼ 0.20, as shown in Figures S8C and S8D , respectively. Analysis of the specific surface area of the produced graphitic carbon is provided in the STAR Methods section, as well as Raman spectra (Figure 13) and time-resolved electrical Joule heating data (Table 2). Tables S1 and S2 provide a summary of all study conditions using the MS and LAS methods, respectively, including I_D/I_G, d₀₀₂, L_c,XRD, temperature, $X_{{\text{CH}}_4}$, $Y_{{\mathrm{H}}_2}$, Y_C, η_ETC, t_res, deposition weight, deposition area, and fiber diameter.

Figure 13.

Effect of vacuum heat treatment on the original carbon cloth properties

Carbon cloth was subjected to Joule heating under vacuum conditions, where Raman characterization (A) reveals localized graphitization domains and (B) retains signatures of the original carbon cloth structure.

Table 2.

Time-resolved electrical measurements of Joule-heated carbon cloth under vacuum conditions

Time (min)	Current (A)	Voltage (V)	Power (W)	Resistance (Ω)
0	–	–	–	–
1	75.7	23.1	1,747	0.305
2	80.6	24.3	1,955	0.301
3	80.6	24.6	1,983	0.305
4	80.6	24.8	2,000	0.308
5	80.5	24.9	2,000	0.309
6	80.6	24.9	2,010	0.309
7	80.5	25.0	2,010	0.310
8	80.5	25.0	2,012	0.310
9	80.5	25.1	2,017	0.311
10	80.5	25.2	2,025	0.313
11	80.5	25.2	2,031	0.313
12	80.5	25.3	2,036	0.314
13	80.5	25.3	2,038	0.315
14	80.5	25.3	2,038	0.315
15	80.5	25.4	2,044	0.315

Moreover, our gas-phase diagnostics and analysis based on species concentrations and balances produce a prediction of solid carbon mass produced (not directly measured because it is capture upstream). We compare the predicted solid carbon masses to the actual measured values, and they are consistently within 5% of each other.

Conclusions

In conclusion, this study presents a new methane pyrolysis method that uses direct Joule heating of the deposition substrate. An electric current passing through a resistive porous carbon cloth generates the necessary heat to break down the hydrocarbon bonds of methane, converting it into valuable graphitic carbon and clean hydrogen fuel. This method leads to the formation of concentric multi-layer graphite around carbon fibers. By examining the effects of power, pressure, flow rate, and reaction time, we report methane conversion rates up to 88%, confirmed by in situ MS and LAS. Characterization of the materials using SEM, Raman spectroscopy, and XRD revealed uniform, high-quality graphite deposition, with an I_D/I_G ratio of 0.1 and 3.38 Å d-spacing under conditions that optimize graphite quality. Additionally, hydrogen-optimized conditions demonstrate a production rate of 2.4 g/h with an electrical-to-chemical efficiency of 2.1%.

This research introduces a new approach to hydrogen production, providing a clean hydrogen fuel source and high-quality graphite. Future work could focus on characterizing the mechanical, thermal, and electrical properties of the graphitic product. Additionally, large-scale production using this methane pyrolysis process can be achieved by implementing continuous processing methods such as roll-to-roll synthesis,^32,60,61 which remains an area for future exploration, as well as studies of long-term stability and avoidance of degradation, enabled, for example, by pulsed heating.⁶² Notably, the voltage and power demands of this process align with photovoltaic solar panel outputs, facilitating direct utilization of renewable solar energy for ohmic heating. Integration of renewable energy enhances the sustainability of this approach, presenting a robust solution to environmental concerns for advanced energy materials.

Finally, we emphasize that unlike typical graphite production methods requiring post-processing at extreme temperatures (>2,000°C), this method enables in situ graphitization at lower energy costs while concurrently producing hydrogen. If lower power or higher flow rates are used, hydrogen production remains high but the carbon could deposit into less-ordered solid forms such as amorphous carbon black, decreasing material value. We believe that these contributions provide a compelling foundation for future work on both application development and process scale-up.

Limitations of the study

While the Joule heating-based methane pyrolysis method demonstrates promising hydrogen and graphite co-production, several limitations remain. The scalability of the process to industrial levels has not yet been validated, particularly with respect to energy efficiency, mass production, and system durability under continuous operation. Determining the appropriate voltage and current range for the power supply can be challenging, as it may vary depending on the type of carbon cloth used—including differences in thickness, grade, and electrical resistivity—as well as dynamic changes in resistivity during the graphitic deposition process. Finally, although high methane conversion rates were achieved, the system’s performance under varying methane purity, the presence of contaminant gases, and real-world gas mixtures requires further investigation.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Timothy Fisher (tsfisher@ucla.edu).

Materials availability

Limited quantities of the graphitic carbon samples produced in this study are available from the lead contact, subject to a completed material transfer agreement.

Data and code availability

Raw data for gas analysis (MS and LAS) and synthesized graphite characterization (Raman and XRD), as well as the full datasets used and/or analyzed during the current study, are available from the corresponding author upon request.

Acknowledgments

The authors thank Mostafa Abuseada for his contribution in co-building the laboratory’s general instrumentation. The authors thank the California NanoSystems Institute at UCLA, United States, and its Elman Family Innovation Fund for support of this work, as well as the US Department of Energy National Energy Technology Laboratory, United States, (DE-FE0032354). A.A. thanks the United Arab Emirates University, United Arab Emirates, for postdoctoral fellowship.

Author contributions

Conceptualization (lead), data curation (equal), formal analysis (equal), investigation (equal), methodology (equal), software (equal), validation (equal), visualization (lead), writing – original draft (lead), writing – review and editing (equal), A.A.; conceptualization (supporting), data curation (equal), formal analysis (equal), investigation (equal), methodology (equal), software (equal), validation (equal), visualization (supporting), writing – original draft (supporting), writing – review and editing (supporting), H.X.; conceptualization (supporting), data curation (supporting), formal analysis (equal), investigation (equal), methodology (supporting), software (supporting), validation (supporting), visualization (supporting), writing – original draft (supporting), writing – review and editing (supporting), B.T.H.; conceptualization (supporting), data curation (equal), formal analysis (equal), investigation (equal), methodology (supporting), software (equal), validation (supporting), visualization (supporting), writing – original draft (supporting), writing – review and editing (supporting), B.J.; conceptualization (supporting), funding acquisition (equal), investigation (supporting), methodology (supporting), project administration (equal), resources (supporting), supervision (equal), writing – original draft (supporting), writing – review and editing (supporting), R.M.S.; conceptualization (supporting), data curation (supporting), funding acquisition (equal), investigation (supporting), methodology (supporting), project administration (equal), resources (lead), supervision (equal), writing – original draft (supporting), writing – review and editing (equal), T.S.F.

Declaration of interests

T.S.F. and R.M.S. are co-founders of SolGrapH Inc., a company specializing in solar-thermal material synthesis. This submitted work is an independent academic study and is not associated with their commercial endeavors or intended as a promotional piece.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work on Overleaf, the authors used TeXGPT/GPT and Writefull in order to assist with minor corrections to grammar, spelling, and sentence structure. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
Tungsten rod	Midwest Tungsten Service Inc.	40300-PureWRod
Tungsten sheet	Midwest Tungsten Service Inc.	42003-WSheet
Radiation shield	Midwest Tungsten Service Inc.	42003-molysheet
Carbon cloth	Fuel Cell Earth LLC	CCP100
Alumina (Al2O3) sheet	Zircar Refractory Composites Inc.	RS-99R
Zirconia (ZrO2) sheet	Zircar Refractory Composites Inc.	NS01-A
Other
Mass flow controller, CH4, 2000 sccm	MKS Inc.	GM50A028103RMM020
Capacitance manometer, 1000 torr	MKS Inc.	624F13TCECB
Vacuum pump	Oerlikon-Leybold Inc.	D65BCS
Exhaust throttle valve	MKS Inc.	T3BIB02K31V0215
Vacuum System Controller	MKS Inc.	946-US-FCFCCM-PC
DC power supply, 3 kW and 30 V	XP POWER LLC	HDS3000PS30
Voltmeter, 0–100 V	Aim Dynamics, Inc.	AIMDC-10V–100V
Current meter, 0–200 A	Phoenix Contact USA, Inc.	MCR-SL-CUC-200-U
PID controller	NATIONAL INSTRUMENTS CORP.	USB-6351

Method details

Gas and pressure control

Pressure was precisely monitored using an MKS 624H capacitance manometer with a full-scale range of 1000 Torr, a resolution of 0.001% of full scale, and an accuracy of 0.12% of the reading. Pressure control was achieved using MKS T3B high-speed exhaust throttle valves, which are designed for applications requiring straightforward yet accurate pressure regulation. These valves offer a control range from 0.0001% to 100% of full scale (when used with dual transducer input) and an accuracy of 0.25% of the set point or 5 mV, whichever is greater.

Flow rates were controlled using calibrated mass flow controllers (MFCs), specifically the MKS GM50A, which features a full-scale range of 40–2000 sccm. The resolution is ±1% of the reading for 20%–100% of full scale, and ±0.2% of full scale for 2%–20% of full scale. While standard MFCs can exhibit reduced precision at low pressures, calibration in this study was performed under conditions closely matching those used in the experiments. This included accounting for potential variations in gas species diffusion through the leak valve and into the RGA compartment.

Mass spectroscopy

A compact, high-resolution residual gas analyzer (RGA) was set up and employed as an in situ mass spectrometer (MS) to detect and measure the composition of the product stream. The RGA (INFICON, TSPTT200) features an ionizer, a quadrupole mass filter, and a 0–200 AMU Faraday cup detector with an electron multiplier, which ensures enhanced peak amplitude and position stability. Given that the RGA must function at pressures below roughly 1 × 10⁻³ Pa, a specialized configuration was designed to allow continuous sampling from significantly higher pressures while keeping the RGA chamber pressure within acceptable limits. This setup included an adjustable leak valve (Kurt Lesker, VZMD9538) attached to the reactor’s outlet stream and connected the RGA compartment to a turbomolecular pump (BOC Edwards, EXT255H/100CF) and a backing rotary vane pump (Edwards, RV3). This arrangement maintains a pressure of about 1 × 10⁻⁵ Pa within the MS chamber. The leak valve can be adjusted to control the sampling from the reactor’s product stream, ensuring that the MS pressure stays within the desired operating range.

Mass spectroscopy (MS) is a powerful tool for quantitatively analyzing gas species, but it can be prone to various errors. These errors stem from non-linear calibration curves over broad ranges, interference between species signals, and inconsistencies in ion detection efficiencies^63,64. Thus, it is essential to perform accurate calibration and use MS in conjunction with other reliable techniques, such as laser absorption spectroscopy (LAS), to achieve precise quantitative results.

The MS was calibrated to yield relative quantitative results (mole fractions) using known gas stream compositions containing the three most anticipated and prominent product species from the methane decomposition process: CH₄, H₂, and C₂H₂. Each gas specie’s composition was regulated with a calibrated mass flow controller under conditions that matched the expected experimental setup. This approach also accounts for potential variations in species diffusion through the leak valve into the RGA compartment. Throughout all calibration runs and MS monitoring tests, the pressure in the RGA chamber was kept at approximately 1 × 10⁻⁵ Pa to minimize sensitivity errors.

For the MS calibration discussed here, H₂ and C₂H₂ were modeled using a first-order approach, while CH₄ was modeled using a second-order approach.³⁴ The methodology initially involved subtracting background from raw current data and identifying the fragmentation patterns (mass spectra) of each species under optimized RGA settings. These fragmentation patterns for the primary species closely matched those found in the NIST database.⁶⁵ For example, the MS recorded fragmentation fractions of H₂ as 2.17% and 97.8% at 1 and 2 AMU, respectively, which compared closely with the NIST database values of 2.06% and 97.94%.

The signal, corrected for background noise, was then used to determine the calibration factors for the main species H₂, CH₄, and C₂H₂ (at 2, 16, and 26 amu, respectively). This process involved applying the fragmentation patterns to the signal while subtracting overlaps among species (e.g., at 26 AMU from C₂H₂, C₂H₄, and C₂H₆) using fragmentation factor data and other adjusted intensities. The resulting values were divided by the calibration factor to obtain adjusted intensities. Each species’ adjusted intensity, along with other intensities, was then used to calculate the mole fraction of that species, ensuring that the calibration factors accurately align with the mole fractions of the gas compositions set by the calibrated mass flow controller.

Calibration factors were determined for H₂ (2 amu), CH₄ (16 amu), C₂H₂ (26 amu), C₂H₄ (28 amu), and C₂H₆ (30 amu) as 0.697, 0.295(1-${\chi }_{{\text{CH}}_4}$) + 0.191, 0.371, 1.83, and 10.6, respectively. The last two calibration factors were derived using fragmentation factors from the NIST database⁶⁵ and fitting MS results to LAS results. This method ensures reliable quantitative results, as mole fractions of C₂H₄ and C₂H₆ do not exceed 2% under various process conditions.

Material characterization

A ZEISS Supra 40VP field emission scanning electron microscope (SEM), equipped with both secondary electron (SE) and backscattered electron (BSE) detectors, was employed to capture SEM images of the carbon substrate. This imaging was completed at an accelerating voltage of 10 kV. In addition, all overview and bundle images were taken at a working distance of 7.8 mm and at magnifications of 100× and 600× respectively. Overview images refer to images of the large-scale structure of the carbon fiber weaves, while bundle images refer to images of individual groupings of unidirectional fibers. When reporting fiber diameters, 5–10 fiber cross-sectional images were averaged. For surface-level elemental mapping, we utilized a Nova Nano 230 SEM with an attached Noran 7 Energy Dispersive Spectroscopy (EDS) system from Thermo Fisher, equipped with a Silicon Drift Detector, operating at an accelerating voltage of 10 kV.

Furthermore, Raman spectra and Raman mapping were obtained using RENISHAW inVia confocal Raman microscope. This microscope operates in conjunction with a 488 nm laser, a 50× magnification objective lens, and a Centrus CCD detector. These tools were used to gather information regarding the peak intensity ratio (I_D/I_G) of the D-peak, representing defects in the lattice, to the G-peak, indicative of in-plane C-C lattice vibrations. The D and G peaks are located at approximately 1350 and 1580 cm⁻¹, respectively, which indicates graphite quality. Additionally, the 2D to G peak ratio (I_2D/I_G), along with the shape and full width at half maximum (FWHM) of the 2D peak situated at approximately 2710 cm⁻¹, provide insights into graphite stacking.

XRD analysis utilized a Panalytical X’Pert Pro instrument with a Cu Kα source (λ_XRD = 1.54° A), operated at 45 kV and 40 mA. Scanning spanned from 5 to 100° (2θ) with a step size of 0.017° and a scan rate of 11°/min. XRD spectra provide structural parameters, including the average interplanar distance (d₀₀₂) from the (002) reflection position (θ₀₀₂) using Bragg’s law³⁷:

$$d_{002}=\frac{{\lambda }_{\text{XRD}}}{2\sin {\theta }_{002}}$$ (Equation 18)

The average crystallite size along the c-axis (L_c,XRD) is determined from the full width at half maximum (FWHM) of the (002) reflection (β₀₀₂) using the Scherrer equation⁶⁶:

$$L_{c,XRD}=\frac{0.9{\lambda }_{XRD}}{{\beta }_{002}\cos {\theta }_{002}}$$ (Equation 19)

To test our hardware setup, we conducted the Joule heating process on a cloth under vacuum conditions. While the overall composition and electrical properties remained largely unchanged, Raman analysis reveals localized signatures of graphitization. Table 2 provides time-resolved electrical measurements of Joule-heated carbon cloth under vacuum conditions. Raman characterization of different regions of the vacuum-processed cloth reveals regions of highly graphitized domains (Figure 13A) and the lower graphitization of the original cloth (Figure 13B).

Specific surface area

The Brunauer–Emmett–Teller (BET) surface area analysis using a Micromeritics 3Flex instrument was conducted on the processed graphitic carbon product under the following conditions: 10 Torr pressure, 30-min duration, 100 sccm methane flow, and 2.5 kW input power (corresponding to an estimated temperature of approximately 2180 K). The analysis revealed a specific surface area of 4.55 ± 0.0086 m²/g, with a BET C constant of 111. After outgassing, the sample mass was 383 mg. The surface area was calculated using adsorption data within the relative pressure range of 0.1–0.3, in accordance with standard BET methodology.

For comparison, the original carbon cloth exhibited a specific surface area of 0.675 ± 0.0036 m²/g, measured over the relative pressure range of 0.05–0.2, with a corresponding BET C constant of 204. The outgassed sample mass in this case was 522 mg. Smooth, macroscopic glass particles were employed as ballast (30.13 g, 13.6 cm³) for both processed and original carbon samples to decrease open volume in the sample bulb, thereby improving accuracy and precision of results.

These results indicate a substantial increase in surface area following graphitic carbon production via methane pyrolysis.

Quantification and statistical analysis

Uncertainty estimates associated with the various measurement techniques are provided in the methodology and results and discussion sections.

Published: September 15, 2025