2.1. Basic Characterization of GOs
The D band and G band are prominent features in the Raman spectra of graphite and graphene materials. The D band is typically located around 1350 cm⁻¹. This band is associated with the breathing modes of sp2 atoms in rings and is indicative of defects or disorder in the graphite lattice. The G band is found around 1580 cm⁻¹. This band arises from the bond stretching of all pairs of sp2 atoms in both rings and chains and is a characteristic feature of graphitic materials.
As shown in
Figure 1a, these bands are used to assess the oxidation degree and structure of carbon-based materials, with the intensity of the D band relative to the G band providing information about the level of disorder or defects present. Due to crystal symmetry, the D band peak at 1350 cm
−1 was quite weak in the original graphite (Gph). The Raman signal intensity indicates the degree of bond cleavage of the graphene oxide lattice and the influence of the edge structure. The intensity ratio I
D/I
G of the two characteristic peaks of GO-80, GO-325, and GO-8000 is 0.8739, 0.9430, and 0.9606, respectively, indicating that the defect density of the three samples increases as the sheet size decreases. GO-8000 shows the highest defect density, which suggests that under the same oxidation conditions, the 8000-mesh graphite has a smaller C–C plane lattice than other graphite during the oxidation process, when the degree of C=C bond cleavage and edge defects is higher, and oxygen-containing functional groups are more likely to be formed.
In the XRD diffraction spectrum (
Figure 1b), there is no (002) diffraction peak at 26.5°, indicating that the graphite is completely oxidized. The characteristic peak positions of (001) of the three GO samples with different mesh sizes are 9.82°, 9.36°, and 9.22°, respectively. The corresponding interlayer spacings are calculated to be 9.01Å, 9.42Å, and 9.57Å, respectively, indicating that the larger the sheet, the smaller the GO layer spacing. The broad peak band at 1627 cm
−1 in the infrared spectrum of the GO was attributed to the skeleton vibration of the benzene ring in unoxidized graphene. The broad absorption peak band around 3400 cm
−1 is the stretching vibration of O–H. The absorption vibration peaks generated by oxygen-containing functional groups at 1398 cm
−1 and 1731 cm
−1 (
Figure 1c) originated from the deformation of –C–O–H and the stretching of –C=O in the –COOH group, respectively. Compared with the vibration peak of the benzene ring, GO-8000 has a higher C=O vibration intensity, which is consistent with the above-mentioned Raman and XRD tests. The absorption vibration peak at 1050 cm
−1 is the vibration of –C–O functional group. The surface of the prepared GO material has abundant oxygen-containing functional groups, and it is expected to show a more sensitive effect on the interaction between the material and water molecules.
XPS is used to identify the existence form and content of carbon and oxygen elements in the GOs. The C1s spectra of GO-80, GO-325, and GO-8000 corresponding to the binding energies of –C–C/–C=C–, –C–O, –C=O, and –O–C=O were found at around 283.8 eV, 285.8 eV, 286.7 eV, and 287.8 eV, respectively (as shown in
Figure 2a–c). The O1s peak centers at 531.5 eV, 532.8 eV, and 533.0 eV were attributed to C–OH, –C–O, and –C=O, respectively (as shown in
Figure 2d–f). The peak shift in the O1s peak position of GO with different meshes is due to the conductivity of GOs. After performing deconvolution fit of XPS peaks, the O1s spectra are obtained (as shown in
Figure 2d–f). Integration of the peaks suggested that most O atoms in the three samples existed in the form of –C–O. In the C1s spectra, the –O–C=O content increased from 1.21% to 3.05%, –C=O increased from 2.64% to 6.12%, and the proportion of –C–O increased from 47.25% to 54.06% as the GO size decreased. In the O1s spectra, the ratio of –C=O increased from 6.22% to 10.11%, the proportion of –C–OH increased from 0.69 to 7.18%, and the proportion of –C–O decreased from 93.08% to 82.71%, as summarized in
Table 1.
As the mesh number increases, the relative content of oxygen element in the sample increases. Among them, the epoxy group located on the GO base plane accounts for the main component, and the carbonyl and carboxyl groups located on the edge account for a smaller proportion. Under the same oxidation conditions, GO with a smaller sheet has more oxygen-containing functional groups.
Graphene oxide is often used in the form of thin films for humidity sensing, so it is necessary to observe the morphology of the thin films. It can be clearly observed from the TEM images (
Figure 3a–c) that the graphite is well exfoliated. The GO dispersion dropped onto the silicon wafer is relatively smooth overall, except for some wrinkles (
Figure 3d–f). The wrinkles on the surface of the films are the essential characteristics of the stable two-dimensional graphene oxides. During the synthesis process of GO, the oxidation of –C=C– bonds lead to the destruction of the conjugated structure and promotes the generation of wrinkles. This phenomenon can simultaneously increase the surface area and enhance the contact efficiency with other molecules, but it will reduce the conductivity of GOs.
2.3. Molecular Dynamics Simulation
In order to clearly simulate the adsorption amount, adsorption energy, and number of interacting hydrogen bonds of water molecules in the three samples, a molecular dynamics simulation system was constructed based on the C and O element contents and the relative content data of each functional group calculated in XPS. The GO multi-layer structure was constructed based on the XRD layer spacing. OPLS-AA (the all-atom optimized potentials for liquid simulations) force field was used to describe H2O molecules, and a large number of water molecules were added to the system and given random movement speeds. For the adsorption process and interaction of water molecules, the NPT (isothermal–isobaric ensemble) integrated system with a step size of 1.0 fs was used to optimize the simulation system. The pressure was set to 1 atm, the temperature was 298.15 K, and a Nose–Hoover controlled thermostat was used. Newton’s equations were used to explain the atomic motion process and the v-Verlet algorithm was used to solve it. LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator, 23 Jun 2022-Update 4) software was used to complete the dynamics simulation process.
When the amount of adsorbed water molecules reaches stability, the adsorption capacity of GO-8000 is 10.71% more than that of GO-80 and 9.53% more than that of GO-325 (
Figure 5a). The adsorption energy between water molecules and oxygen-containing functional groups is a key aspect of intermolecular interactions. This interaction is mainly formed through hydrogen bonds, and the strength and stability of hydrogen bonds directly affect the ability of water molecules to combine with oxygen-containing functional groups. The absorption energy of water molecules on the surface of GOs is shown in
Figure 5b. The higher the negative value of these data, the stronger the interaction. Then, the order of interaction strength is GO-8000 > GO-80 > GO-325 accordingly. The results were further confirmed by the number of hydrogen bonds, as shown in
Figure 5c. Due to the difference in the relative content of oxygen-containing functional groups in the prepared materials, the most hydrogen bonds were found in GO-8000, followed by GO-325, and GO-80 had the least hydrogen bonds.
The adsorbed water molecules are distributed closely to the functional groups, showing a higher peak relative to the center of the slit. Moreover, this part of the water molecules overlaps with the functional groups at the plane, indicating that a small number of water molecules fill the gaps between the functional groups, and the part of water molecules distributed close to the plane is greatly affected by the plane and the functional groups. The distribution characteristics of water molecules in different slits have certain regularity. The peak heights of the water molecule peaks in the slit system are 0.9~1.1 g/cm
3, 1.0~1.2 g/cm
3, and 1.1~1.3 g/cm
3 (
Figure 5d), respectively. The arrangement of water molecules in the slit suggested characteristics different from those of bulk water molecules. Planes and functional groups have a greater influence on the water molecules in the slit.
2.4. Humidity Response Performance
Graphene oxide itself is not a typically p-type or n-type material. In its natural state, graphene oxide may exhibit a certain p-type behavior. This is because the oxygen-containing functional groups on the surface can serve as electron acceptors and extract electrons from the graphene plane, leaving more holes, resulting in p-type carrier-dominated conductivity. The migration of electrons and holes has an important impact on humidity sensing.
The three sample dispersions were drop-coated onto interdigitated electrodes, and a micro-voltage was applied to test the humidity sensing performance at 15~92% RH (
Figure 6a–c). As the humidity increases, the resistance decreases until a certain humidity condition is reached. The adsorbed water molecules can accept electrons from GO to form hydrated electrons. This electron transfer will cause changes in the electronic structure of GO, thereby affecting its conductivity. In addition to the movement of electrons, the adsorption of water molecules may also lead to the generation of holes. In GO, the presence of water molecules can prompt oxygen atoms to release electrons to adjacent carbon atoms, leaving holes at the oxygen atoms. Such an increase in holes will also change the conductivity of GO. Due to the migration of electrons and holes, the conductivity of GO changes with changes in humidity. Under low-humidity conditions, the conductivity of GO is low because there are not enough free electrons and the conductive path is clear. Under high-humidity conditions, the conductivity of GO will increase due to the occupation of electrons by water molecules and the increase in the number of holes.
The electrochemical stability of the three electrodes was tested by switching RH between 15% and 92% (
Figure 6d–f). GO-80 showed the smallest resistance, the slowest response speed, and the fastest recovery time during the test process. The GO-8000 sample had the largest resistance value, the fastest response speed, and the slowest recovery speed. The response times and recovery times of the three samples are 6 s, 3 s, and 2 s and 12 s, 28 s, and 48 s, respectively. The desorption time of water molecules on the sample film is much longer than its adsorption time. This is because, during the process from adsorption to complete desorption of water molecules, the molecules need to obtain enough energy to overcome the attraction between adjacent molecules and thus be able to detach from the sample. For water molecules that enter between the GO sheets and undergo hydration reactions, the water molecules easily form larger clusters due to hydrogen bonds, and the desorption process often requires more energy and time. The desorption process lags behind the adsorption process and the adsorption plots are consistent with the type IV isotherm, indicating single-layer and multi-layer adsorption on micro-mesoporous adsorption materials. As the humidity increases, the sample resistance continues to decrease, and the curve becomes flat after reaching a certain humidity condition.
The corresponding maximum resistances of the three samples are 1.82 × 10
7 Ω, 1.43 × 10
8 Ω, and 1.53 × 10
8 Ω, respectively. When the ambient humidity reaches 92%, the corresponding resistance values of the samples are 5.41 × 10
5 Ω, 1.87 × 10
6 Ω, and 2.32 × 10
6 Ω, respectively (
Figure 7a–c). Linear fitting was performed on the humidity response sensitivity changes of the resistance of the samples in the humidity ranges of 15% to 57% RH and 57 to 92% RH. It was found that when the humidity reached about 57%, the decline curves of the resistance values of the three samples all slowed down significantly (
Figure 7d).
The illustration of sensitivity change curves of GO-80, GO-325, and GO-8000 in response to humidity is shown in
Figure 8. Linear fitting was performed for the humidity response sensitivities of 15~57% RH and 57~92% RH, respectively.
R0 is the resistance of the sample under initial conditions (15% RH, room temperature), Ri is the resistance under the corresponding humidity, and ΔRH is the change in humidity relative to the initial conditions. Rp is the sample response sensitivity.
When the humidity changes from 15% to 23%, the corresponding sensitivities of the samples are 26.37%, 32.38, and 33.99%, respectively. The humidity response sensitivity shows an obvious gradient phenomenon, and as the humidity increases, the response sensitivity of GO-8000 is higher than that of GO-325 and GO-80 in the RH range of 15% to 57%. When the humidity reaches 92%, GO-8000 shows lower response sensitivity than GO-325 and GO-80. At this time, the corresponding sensitivities of the three samples are, respectively, 97.03%, 98.69%, and 98.48%.
2.5. Humidity Response Mechanism
The introduction of oxygen-containing functional groups into graphene materials has different effects on their conductivity. The hydroxyl groups in GO-8000 can form hydrogen bonds with the surface of graphene, thereby hindering charge transport and thus reducing the conductivity of the material to a certain extent. The most hydrogen bonds were generated in GO-8000 (
Figure 5c). The presence of carboxyl groups has a more significant impact on the conductivity of GO. The XPS results of both C1s and O1s show that GO-8000 has a large number of carboxyl groups. Carboxyl groups can interact with the graphene surface and change its electronic structure. In addition, the dissociation of carboxyl groups may lead to ion transport, further reducing the conductivity (
Figure 7d). Epoxy groups can form covalent bonds with the graphene surface, destroying the conjugated structure of graphene, thereby affecting its conductivity, such as GO-80 samples.
At low humidity (RH < 15%), water molecules enter GO through diffusion and interact with surface oxygen-containing functional groups and tend to be between GO sheets, and GO is in a low-hydration state with less H
+ and free electrons. There are many ion barriers, which hinder the movement of ions and have high resistance. When the humidity gradually increases, the role of hydroxyl and carboxyl groups is to increase the occupation of electrons by water molecules and increase the number of holes, and the mechanism of water adsorption begins to change. The adsorbed water molecules that have diffused into the inside of the sheets quickly reduce the drag by increasing the hydronium ions between the GO sheets. After RH > 57%, the concentration of water molecules reaches a critical value. Limited by the number of oxygen-containing functional groups, the number of hydrated ions reaches the upper limit. At this time, the resistance reduction mainly depends on the change in the content of adsorbed water in the layer. Moreover, when the adsorbed water content reaches a certain level, water aggregates will form within the sheet, hindering subsequent water adsorption and slowing down the resistance reduction rate (
Figure 7b,c). Therefore, the large number of carboxyl groups in GO-8000 will hinder further intercalation of water molecules and exhibit lower sensitivity (
Figure 8).