Understanding the Detection Mechanisms and Ability of Molecular Hydrogen on Three-Dimensional Bicontinuous Nanoporous Reduced Graphene Oxide

Abstract

Environmental safety has become increasingly important with respect to hydrogen use in society. Monitoring techniques for explosive gaseous hydrogen are essential to ensure safety in sustainable hydrogen utilization. Here, we reveal molecular hydrogen detection mechanisms with monolithic three-dimensional nanoporous reduced graphene oxide under gaseous hydrogen flow and at room temperature. Nanoporous reduced graphene oxide significantly increased molecular hydrogen physisorption without the need to employ catalytic metals or heating. This can be explained by the significantly increased surface area in comparison to two-dimensional graphene sheets and conventional reduced graphene oxide flakes. Using this large surface area, molecular hydrogen adsorption behaviors were accurately observed. In particular, we found that the electrical resistance firstly decreased and then gradually increased with higher gaseous hydrogen concentrations. The resistance decrease was due to charge transfer from the molecular hydrogen to the reduced graphene oxide at adsorbed molecular hydrogen concentrations lower than 2.8 ppm; conversely, the resistance increase was a result of Coulomb scattering effects at adsorbed molecular hydrogen concentrations exceeding 5.0 ppm, as supported by density functional theory. These findings not only provide the detailed adsorption mechanisms of molecular hydrogen, but also advance the development of catalyst-free non-heated physisorption-type molecular detection devices.

1. Introduction

Detection of molecular hydrogen in ambient atmospheres has become increasingly important because renewable energy, especially hydrogen-based energy, is a focus for the development of sustainable societies [1,2]. Users may encounter some risks associated with the initial leakage of molecular hydrogen from equipment, and explosions can occur with the deployment of molecular hydrogen equipment such as fuel cells. Therefore, the identification of an accurate detection mechanism for molecular hydrogen under hydrogen flow conditions is urgent for the further development of hydrogen gas sensors and the environmental safety of societies that utilize hydrogen.

Conventional molecules are detected by catalytic combustion, resistance change, and thermal conductivity change [3,4,5,6,7]. Detection is generally accompanied by chemical reactions on the detection element. Heating of the elements may be required for molecular detection. However, such systems might lead to undesirable situations such as spark generation, irreversible exothermic reactions, or overheating, which could occur under abnormal operating conditions in uncontrolled devices. Thus, the possible hazards related to hydrogen gas combustion in conventional gas detection devices cannot be completely excluded. Recently, non-heated and non-chemical reaction detection mechanisms have been reported, employing monolayer or multilayer graphene. These graphene sensors successfully detected gaseous polar molecules such as H₂O, CO, CO₂, NO, NO₂, O₂, SO₂, and NH₃ at parts per billion (ppb) and parts per million (ppm) levels [8,9,10,11,12,13]. Moreover, reduced graphene oxide (rGO) flakes demonstrated highly sensitive polar molecule detection without any catalytic metals [14,15,16,17,18,19], because the molecular adsorption sites of hydroxyl (C-OH), epoxide (C-O-C), ketone (C=O), and carboxyl (COOH) groups on rGO are effective physisorption sites. However, the detection mechanism of non-polar molecules such as molecular hydrogen (H₂) using either graphene or rGO flakes is not precisely discussed in the literature because the weak physisorption strength and insufficient molecular detection area of the graphene surface mask the details of the electrical resistance behaviors. Since catalytic H₂ detection with Pt [20,21], Pd [22,23,24,25], Au [26], ZnO [27,28], and SnO₂ [29,30] at high temperatures has been predominantly employed for detecting H₂, the significant challenge of realizing physisorption of non-polar H₂ and revealing the detection mechanisms in detail remains in the development of non-catalytic H₂ detection devices.

Here, we synthesized a single and monolithic nanoporous reduced graphene oxide (np-rGO) sheet. The synthesized sheet had an edge-free bicontinuous and open porous three-dimensional (3D) structure and was treated by a modified Hummer method. The np-rGO sheet maintained a projected surface area 500 times larger than that of a standard two-dimensional (2D) graphene sheet, performing significantly accurate detection via molecular physisorption of H₂ as compared to pristine nanoporous unoxidized graphene and rGO flakes. Such physisorption detection mechanisms based on a large surface area provide a new class of design direction for non-heated, flow-type, and physisorption-type graphene-based detection systems for non-polar gas molecules without metal catalysts.

2. Materials and Methods

2.1. Preparation of Nanoporous Ni Substrates

Ni₃₀Mn₇₀ ingots were prepared by melting pure Ni and Mn (purity >99.9 at.%) in an Ar-protected arc melting furnace. After annealing at 900 °C for 24 h for microstructure and composition homogenization, the ingots were cold-rolled into thin sheets (50 μm) at room temperature. Nanoporous Ni was prepared by chemical dealloying in a 1.0 M (NH₄)₂SO₄ aqueous solution at 50 °C overnight. After dealloying, the samples were rinsed thoroughly with distilled water and then rinsed with ethanol before vacuum drying.

2.2. Preparation of Nanoporous Graphene

The nanoporous Ni substrates were loaded in a quartz tube (Ø26 × Ø22 × 250 mm), placed in a quartz tube furnace (Ø30 × Ø27 × 1000 mm), and annealed at 800 °C under a mixed gas flow (200 sccm Ar and 100 sccm H₂) for 3 min [31,32,33,34,35]. After the reduction pre-treatment, a mixed atmosphere of H₂ (100 sccm), Ar (200 sccm), and benzene (0.5 mbar, 99.8%, anhydrous, Aldrich) was introduced for graphene growth at 800 °C for 2 min. Subsequently, the furnace was opened for the rapid cooling of the inner quartz tube. The nanoporous Ni substrate was etched in a 1.0 M HCl solution and then the remaining nanoporous graphene was transferred into a 2.0 M HCl solution to completely remove residual Ni. The samples were repeatedly washed with water and subsequently kept in water.

2.3. Preparation of Nanoporous Graphene Oxide and Nanoporous Reduced Graphene Oxide

The nanoporous graphene dispersed in water was oxidized via a modified Hummer method [36]. A single graphene sheet was added to a KMnO₄ (0.45 g), concentrated H₂SO₄ (3.45 mL), and NaNO₂ (0.075 g) mixture without stirring. After 10 min, the mixture was diluted in an aqueous H₂SO₄ (5%) solution. Then, an aqueous H₂O₂ (30%) solution diluted in aqueous H₂SO₄ (5%) was slowly added to the mixture. The obtained transparent sheet was then washed several times with a mixture of aqueous H₂SO₄ (3%) and H₂O₂ (0.5%). The nanoporous graphene oxide (GO) sheet was then repeatedly washed with water and kept in 2-propanol. The samples were dried by a conventional supercritical CO₂ drying method. The dried GO sample was transferred onto a SiO₂ substrate and then connected to fine Cu lines (twisted) using silver paste in a standard pseudo four-probe configuration. The reduced graphene oxide (rGO) sample was prepared by annealing the GO sample on the substrate at 150 °C using ribbon heaters under vacuum in a self-made atmosphere-controlled glass cell connected to Swagelok vacuum components.

2.4. Supercritical Drying Procedure

To prevent the collapse of the fragile nanoporous structures caused by water capillary forces during drying, the nanoporous graphene and rGO samples immersed in 2-propanol were dried by using supercritical CO₂ (scCO₂) and thus 2-propanol was substituted for CO₂ without the action of capillary forces. To this end, the nanoporous graphene sample was firstly transferred to a vial (volume: 5 mL) filled with 2-propanol (400 μL), and then the bottle was placed in an 80-mL pressure-resistant container (TAIATSU techno Corp). After removing the air inside the container via CO₂ purging, the pressure of the container was gradually increased to 15 MPa by introducing liquid CO₂ (5 Mpa, −4 °C, density: 0.964 g mL⁻¹) at a flow rate of 20 mL min⁻¹ (~19 g min⁻¹) using a high-pressure plunger pump (NP-KX-540, NIHON SEIMITSU KAGAKU Co. Ltd.). The scCO₂ drying process was performed at 70 °C and at a constant CO₂ flow rate of 5 mL min⁻¹ (~4.8 g min⁻¹) by forming a homogeneous phase of 2-propanol and scCO₂, thus minimizing the capillary force. The pressure was maintained at 15 MPa for 5 h during the drying process. After drying the sample completely, the temperature was set at 40 °C and the container was gradually depressurized over 43 h from 15 MPa to atmospheric pressure by gradually pumping CO₂ from the system.

3. Microstructure Characterization and Property Measurements

3.1. Imaging and Spectroscopic Characterization

The microstructures of the nanoporous GO and nanoporous rGO were characterized using a scanning electron microscope (SEM; JEOL JSM-6700). Chemical analyses on the graphene samples were performed via X-ray photoelectron spectroscopy (XPS; AXIS ultra DLD, Shimazu) with an Al Ka X-ray monochromator. Raman spectra were obtained using a micro-Raman spectrometer (Renishaw InVia Reflex 532) with an incident wavelength of 532.5 nm. The laser power was set at 0.1 mW to avoid possible damage or unexpected reduction by laser irradiation. The graphene samples were placed on a background-free glass slide. The accumulation time of each spectrum was 100 s.

3.2. Transport Property Measurements

The temperature and magnetic field dependences of the electrical resistance of reduced nanoporous graphene oxide were measured using the four-probe method under a magnetic field at 300 K. The reduced nanoporous graphene sheets were placed on a SiO₂/Si (SiO₂: 300 nm) substrate. The electrode was fabricated using Ag epoxy (H20E, Epoxy Technology). The electrical resistance measurements were performed using the Physical Properties Measurement System (PPMS; Quantum Design).

3.3. Gas Detection Device Fabrication and Measurements

A molecular detection glass cell was fabricated using an atmosphere-controlled separable three-neck glass cell with an O-ring. Swagelok vacuum components were inserted into two rubber stoppers, and the stoppers were installed in two of the necks. The Swagelok vacuum components were connected to mass flow controllers and a rotary pump. Four Cu wires (1.5 mm) and a thermocouple (type K) were inserted in a rubber stopper, and the stopper was installed in the third neck of the cell to communicate the inside and outside of the cell and the device; all connections were completely sealed with Torr Seal^® to avoid leakage. The detection element of the nanoporous GO sheet (7.2 mm × 2.5 mm) was attached on a SiO₂ substrate through fine twisted Cu lines fixed with silver paste. Then, the substrate was transferred to the glass cell and the fine Cu lines were electrically connected to the Cu wires (1.5 mm) inside the cell. The temperature inside the cell was controlled by an external ribbon heater for thermal reduction of nanoporous GO in the cell. The whole system and the connected vacuum components were heated to 150 °C, and the GO (already connected to the Cu wires (1.5 mm) through the fine Cu wires) was thermally reduced to rGO under vacuum (5 Pa). Briefly, the temperature for thermal reduction was fixed at 150 °C and the heating time was set at 4 h for oxygen contents of 20.0 at.%, 6 h for oxygen contents of 18.2 at.%, and 12 h for oxygen contents of 10.8 at.%. After cooling down to the room temperature (23 °C), the thermally reduced GO samples were sequentially measured in the cell by a standard pseudo four-probe method using a semiconductor parameter analyzer (Keysight B1500A). The applied voltage was set at −9 V for the nanoporous rGO samples (with chemical oxidation and thermal reduction treatments) and to −10 mV for the pristine nanoporous graphene (without oxidation and reduction treatments) under a mixture of gaseous argon (Ar: >99.9999 vol.%, H₂O: <−80 °C, O₂: <0.1 ppm, CH₄: <0.1 ppm, CO₂: <0.1 ppm, CO: <0.1 ppm, N₂: <0.6 ppm, H₂: <0.1 ppm). The ultrapure gaseous Ar continuously flowed until the resistance value stabilized to exclude unexpected influences of impurities such as H₂O, O₂, CH₄, CO₂, CO, and H₂. Thus, any changes in resistance values under the mixture of H₂/Ar gas flowing could be attributed to the introduction of ultrapure H₂ gas. After the resistance was stable, the mixture of gaseous Ar and gaseous hydrogen (H₂: >99.9999 vol.%, H₂O: <−80 °C, O₂: <0.1 ppm, CH₄: <0.1 ppm, CO₂: <0.1 ppm, CO: <0.1 ppm) was introduced in the cell. The total flow rate of the mixed gas was set at 30 sccm. To obtain the 1 ppm H₂ gas in the Ar flow case, gaseous hydrogen was mixed with gaseous argon to a H₂ concentration of 0.0001 vol.%, and then the mixed gas was supplied by the mass flow controllers at a flow rate of 30 sccm (the partial pressure of hydrogen was 0.10 Pa). Note that flow rates exceeding 30 sccm deformed the atomically thin porous structures, and the resistance values became unstable. Before all measurements, the whole system was vacuumed for 12 h at room temperature to avoid further thermal reduction of the rGO.

4. Density Functional Theory Calculations

The geometries of the surface models (C₅₄H₁₈, C₅₄H₁₈-2OH, C₅₄H₁₈-2O, C₅₃H₁₈-COOH, and C₅₃H₁₈-2O) were optimized using the density functional theory (DFT) method with the B3LYP functional [37,38] and considering the D3 version on Grimme’s dispersion correction [39] with 6-311G(d,p) basis sets [40]. For the graphene oxide models, the coordinates of the terminal H atoms were fixed to those of the pristine graphene to model a large surface that would prevent distortion. Several structures from the graphene oxide models were compared and the most stable alternative was chosen for the calculation of H₂ adsorption. All calculations were performed using Gaussian 16 quantum chemistry software [41].

5. Results

5.1. Material Characterization

The transparent 3D np-GO (Figure 1a) was dried by a conventional supercritical CO₂ fluid method, attached to a SiO₂ substrate with the Cu wires (Figure 1b), and transferred to the atmosphere-controlled glass cell to reduce the np-GO sample at 150 °C for 4 h under vacuum. The dried np-rGO maintained the original morphology of the pristine 3D nanoporous graphene without any obvious flakes or holes after the oxidation and reduction processes (Figure 2a and Figure S1). The specific surface area and average pore size were ~1000 m² g⁻¹ and ~200 nm, respectively, and the total projected surface area exceeded 500 times that of the 2D graphene sheet [32,36,42]. Moreover, it was determined by high-resolution transmission electron microscopy that amorphous-like graphene lattices on np-rGO were partially crystallized after thermal reduction, and electron energy-loss spectroscopy (EELS) spectra also demonstrated regeneration of sp² bonds by post reduction [36]. EELS mappings confirmed that the oxidation species were almost homogeneously distributed on the nanoporous structures [36].

Raman spectra were analyzed to estimate the quality of the graphene samples and their respective oxidation levels (Figure 2b and Figure S2 and Table S1). The pristine nanoporous graphene showed an intensity ratio of the 2D and G bands (I_2D/I_G) of 2.5 and of the D and G bands (I_D/I_G) of 0.08, corresponding to high-quality 1–2 layer graphene without detectable amorphous carbon features [43,44]. After the oxidation process, the np-GO showed a large defect density (I_D/I_G of 1.6 and I_2D/I_G of 0.91). The np-rGO partially restored graphene features (I_D/I_G of 0.98 and an I_2D/I_G of 1.7). The full width at half maximum (FWHM) values of the D, G, and 2D bands of the np-rGO were significantly increased in comparison to the pristine nanoporous graphene. Chemical binding was further investigated by XPS. The C1s X-ray photoelectron spectrum (Figure 2c,d, Figures S3 and S4) demonstrated an oxygen content of 4.7 at.% for pristine nanoporous graphene, 32.8 at.% for np-GO, and 20.0 at.% for np-rGO. Oxidized carbon species such as 10.7 at.% C-O and/or C-O-C (286.6 eV), 8.7 at.% C=O (288.8 eV), and 0.6 at.% C(O)OH (290 eV) were observed on the np-rGO, which was consistent with reported observations [36,45,46,47]. The FWHM of graphitic carbon at 284.5 eV was broadened from 0.71 eV of pristine nanoporous graphene to 2.0 eV of np-GO after oxidation and narrowed to 1.3 eV of np-rGO after reduction; this was associated with the mixed states of orbitals sp² and sp³ [46,47]. Therefore, the np-rGO presented a significant number of defect structures containing C-O, C-O-C, and C=O with a small number of C(O)OH. Moreover, it was confirmed that the X-ray photoelectron spectra of Ni 2p and Mn 2p were not detected in any of the samples (Figures S3 and S4), which was consistent with the absence of Ni and Mn compounds in the X-ray diffraction (XRD) spectrum (Figure S5). Other np-rGO samples with different oxygen contents were similarly characterized (Figures S2–S4). Furthermore, the inter-defect distance on the graphene lattice estimated by Raman spectra and the distance between functional groups estimated by XPS were compared. The average inter-defect distance was estimated by the I_D/I_G from the Tuinstra–Koenig relation: I_D/I_G = C(λ)/L_D, where C(λ) is a proportionality constant at the excitation laser wavelength λ and L_D is the average distance between the defects [48,49,50]. Under the assumption that this relation could be applied to the rGO system, the L_D value of np-rGO with an oxygen content of 20 at.% (I_D/I_G = 0.98) was approximately 8–10 nm, which was almost of the same order as the distance between the oxidized functional groups estimated by the atomic concentrations.

The transport properties of nanoporous graphene and np-rGO were measured at 300 K by a standard four-probe method. The typical electrical conductances and electron mobilities of the nanoporous graphene samples were 500–1000 S cm⁻¹ and 5000–8000 cm² V⁻¹ s⁻¹, respectively [35,51]. The typical electrical conductances and electron mobilities of the np-rGO samples with degrees of oxidation of 19.6–23.3 at.% were 0.0025–0.043 S cm⁻¹ and 0.31–1.0 cm² V⁻¹ s⁻¹, respectively [36]. Note that the electron mobility of the np-rGO samples was ca. 2 orders of magnitude higher than that of conventional rGO films formed by small flakes under the same thermal reduction conditions (0.0015–0.06 cm² V⁻¹ s⁻¹) [52]. Therefore, the electrons on np-rGO were moved smoothly owing to the unique bicontinuous structures.

5.2. Hydrogen Adsorption

To evaluate the molecular physisorption ability of H₂, we measured the time dependence of resistance changes under the H₂/Ar mixed gas in the atmosphere-controlled glass cell. Firstly, we confirmed the reversibility of molecular physisorption of H₂ under Ar and H₂/Ar flow conditions (500 ppm H₂, Figure 3a) on the np-rGO with an oxygen content of 20.0 at.%. In the first step, pure Ar gas flowed until the resistance value became constant, and then H₂/Ar gas flowed; time was necessary to replace the pure Ar gas in the cell with H₂/Ar gas. The resistance values gradually increased as the H₂ reached the surface of the np-rGO. The resistance values kept substantially constant at the equilibrium states of adsorption and desorption of H₂ under the H₂/Ar gas flow. The resistance values gradually decreased to the initial resistance value after stopping the H₂ gas supply (i.e., 30 sccm pure Ar gas flow only). The response was relatively slower than that of conventional catalytic metal-based gas detection systems, probably because the replacement of Ar by H₂ requires time owing to the low flow rate and geometrically requires time to reach the surface due to the complicated porous structure [32] in comparison to the flat structures [20,21,22,23,24,25,26,27,28,29,30].

The change in resistance at H₂ gas volume concentrations from 0.1 ppm to 10,000 ppm under H₂/Ar gas flow (30 sccm in total) was investigated for the np-rGO with an oxygen content of 20.0 at.% (Figure 3b and Figure S6) and nanoporous graphene (Figure 3c). We found that the time-dependent resistance values at −9.0 V of the np-rGO (initial resistance: 742 MΩ) increased above 5 ppm H₂ under H₂/Ar gas flow, whereas they decreased below 2.8 ppm H₂. Each change of resistance value was normalized to the equilibrium state before H₂ adsorption (only Ar flow state) and plotted for the various H₂ ppm values (Figure 3d). The normalized changes (%) were proportional to logarithmic H₂ ppm values at above 5 ppm H₂ under H₂/Ar gas flow. The nanoporous graphene (initial resistance: 33.8 Ω) showed a similar behavior but did not show large changes in resistance values for 0.1–10,000 ppm H₂ under the same conditions. Thus, the large changes in np-rGO were attributed to the oxidized functional groups on the graphene. Moreover, both samples exhibited a local minimum at approximately 1 ppm H₂ under H₂/Ar gas flow. Note that these increases and decreases have not been reported for graphene and rGO detection systems [8,14,15,16,17,18,19]. Furthermore, the oxidization level dependence of the resistance changes was similarly investigated (Figure S7). The np-rGO samples with oxygen contents of 10.8 at.% (initial resistance: 11.1 kΩ) and 18.2 at.% (initial resistance: 1.26 MΩ) showed similar behaviors, but smaller resistance changes at typical H₂ ppm values under H₂/Ar gas flow in comparison to the np-rGO sample with an oxygen content of 20.0 at.%.

5.3. DFT Calculations

To elucidate the resistance behaviors, we performed DFT calculations for molecular physisorption of H₂ on the surface of graphene and graphene oxide (Figure 4a–d and Figure S8 and

Table 1. Summary of the highest occupied molecular orbital (HOMO) energy level, the lowest unoccupied molecular orbital (LUMO) energy level, the shift of HOMO (ΔE_HOMO) and LUMO (ΔE_LUMO) energy level and cluster total energy before and after adsorption of molecular hydrogen. Charge transfer from H₂ to GO indicates the sum of the Mulliken charge of the GO model.

	Before Absorption of H2		After Absorption of H2
	HOMO (eV)	LUMO (eV)	ΔEHOMO (eV)	ΔELUMO (eV)	ΔEabs (kcal/mol)	Charge Transfer from H2 to GO
C54H18	−0.19069	−0.08710	−0.0002	−0.0001	−1.58	−0.007 e
C54H18-2OH	−0.19323	−0.09767	−0.0004	−0.0007	−3.14	+0.023 e
C54H18-2O	−0.19819	−0.09185	−0.0004	−0.0005	−2.73	+0.020 e
C53H18-COOH	−0.17395	−0.12959	−0.0003	−0.0005	−2.72	+0.018 e
C53H18-2O	−0.18408	−0.09939	−0.0003	−0.0007	−2.88	+0.019 e

). We employed a nanographene cluster model with C₅₄H₁₈ and partly oxidized the nanographene to introduce oxygen-containing functional groups such as hydroxyl, epoxide, ether, ketone, and carboxyl groups for the adsorption of a single H₂ molecule (Figure S9). The adsorption energy of H₂ was calculated in these models and was −3.14 kcal/mol for C₅₄H₁₈ with two hydroxyl groups, −2.73 kcal/mol for C₅₄H₁₈ with two epoxide groups, −2.72 kcal/mol for C₅₄H₁₈ with a carboxyl group, and −2.88 kcal/mol for C₅₃H₁₈ with an ether and a ketone group; these values were lower than that for C₅₄H₁₈ without any functional groups (i.e., −1.58 kcal/mol), and there were no significant differences in the functional groups. In the models, it was found that two kinds of charge transfer phenomena occurred: (1) from oxygen-containing functional groups to graphene oxide (i.e., hole doping initially occurred from oxygen-containing functional groups to graphene) and (2) from a single H₂ molecule to graphene oxide with +0.020 e (i.e., additional hole doping occurred from H₂ to graphene) (Figure 4 and Figure S10 and Table 1). This suggests that the np-rGO before adsorption was electron-deficient because of the oxidized functional groups and became further electron-deficient after H₂ adsorption. Moreover, the energy levels of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) decreased after the adsorption (Table 1). In particular, the energy level of the LUMO of the C₅₄H₁₈ with two hydroxyl groups showed the highest reduction, in which the adsorption of H₂ modified the electronic state of the rGO, benefitting the sequential adsorption of additional H₂. Notably, we found a sequential adsorption of three additional H₂ molecules by the C₅₄H₁₈ model with two hydroxyl groups (−10.90 kcal/mol in total) and a charge transfer (+0.074 e by four H₂ molecules) from the additional H₂ molecule on graphene oxide further occurred (Figure S11 and Table S2).

6. Discussion

The concentrations of adsorbed H₂ lower than 1 ppm showed negative resistance changes, while those higher than 5 ppm showed positive resistance changes. The high resistance changes are attributed to the high oxidation level of the samples (Figure S7) and lower than 1 ppm or higher than 500 ppm H₂ concentrations (Figure 3 and Figure S7). This suggests that low amounts of adsorbed H₂ should promote the charge transfer effect (Table 1), while high amounts of adsorbed H₂ should present a larger influence on resistance change and promote an additional charge transfer effect on the oxidized functional groups (Table S2). Combining the experimental results and DFT calculations, two mechanisms of resistance change could arise for the threshold amount of 1–5 ppm H₂ under H₂/Ar gas flow (30 sccm in total). Below 1 ppm H₂, contribution of Coulomb scattering by the adsorbed H₂ was not dominant/apparent and the charge transfer from H₂ to np-rGO was a major effect. In this case, further hole doping from the adsorbed H₂ of the p-type np-rGO reduced the resistance (Figure 4e). Conversely, above 5 ppm H₂, Coulomb scattering was the dominant effect, and charge transfer from H₂ to np-rGO was a minor effect. The resistance was simply enhanced by the increase of charged impurity centers (Figure 4e,f), in accordance with a previous report [9]. Notably, even for 5 ppm H₂, it was observed that the resistance firstly decreased from 0 s to 300 s (charge transfer domination) and gradually increased after 300 s (Coulomb scattering domination) with an increase in hydrogen concentration (Figure S12). Thus, in the range from 1 to 5 ppm H₂, the Coulomb scattering and charge transfer effects may be comparable, giving rise to a local minimum. Moreover, the DFT calculations suggest that the hydroxyl groups play an important role in the sequential adsorption of a large amount of H₂. The np-rGO with an oxygen content of 20.0 at.% (hydroxyl/epoxide group: 10.7 at.%) showed the largest change of resistance values at higher H₂ ppm values in comparison to the nanoporous graphene and np-rGO with oxygen contents of 10.8 at.% (hydroxyl/epoxide group: 2.5 at.%) and 18.2 at.% (hydroxyl/epoxide group: 6.3 at.%). Furthermore, open porous structures with large total surface areas whose surface was modified by the oxygen-containing functional groups significantly promoted mass transfer and increased the sensitivity of molecular physisorption of non-polar H₂. The excellent transport properties resulting from such monolithic characteristics supported accurate detection of resistance changes.

7. Conclusions

We developed a novel 3D nanoporous rGO for molecular hydrogen detection at room temperature under gas flow conditions. The detection mechanisms of molecular hydrogen were revealed by using 3D nanoporous rGO specimens featuring a large surface area, a single bicontinuous monolithic sheet morphology, and unique transport properties (non-electron hopping conduction), and were found to be based on the charge transfer effect and Coulomb scattering effect. Moreover, DFT calculations revealed that hydroxyl, epoxide, ether, ketone, and carboxyl groups are preferable molecular physisorption sites, and that hydroxyl groups benefit from the sequential adsorption of molecular hydrogen. Nanoporous rGO with tunable oxygen-containing functional groups could provide a general platform for a room-temperature physisorption-type gas sensor and provide a new design direction for nanoporous material-based gas sensors for combustible gases.