Exploration of the temperature-dependent correlations present in the structural, morphological and electrical properties of thermally reduced free-standing graphene oxide papers

The raw materials needed for the preparation of GO suspension included graphite powder (99% purity), hydrogen peroxide (H₂O₂, 30%) obtained from Sigma-Aldrich (USA) and potassium permanganate (KMnO₄), hydrochloric acid (HCl, 37%) and sulfuric acid (H₂SO₄, 98%) purchased from Merck (Germany). Whatman™ cellulose nitrate filter papers with 0.2 µm pore size and 47 mm diameter were also purchased from Merck (Germany).

Firstly, a GO suspension was prepared following the process previously described by Phrompet et al [78]. In short, graphite powder was oxidized using a modified Hummers’ method to obtain graphite oxide. The resulting yellowish-brown product was separated by centrifugation followed by washing in HCl to remove sulfates. The obtained solution was then filtered and washed several times until a solid paste with pH value of approximately 7 was obtained, which was then dried in an oven at 65 °C overnight to obtained powdered graphite oxide. The powder was subsequently re-dispersed in distilled water followed by ultrasonication and centrifugation to finally obtain a dark brown supernatant of aqueous suspension of GO nanosheets with a concentration of 5 mg/ml.

The high dispersion qualities of GO in water makes it a suitable candidate for vacuum filtration. About 2 ml of the aqueous solution of GO was poured through a cellulose nitrate filter paper (0.2 microns pore size) fitted to a typical vacuum filtration apparatus shown in Fig. 1b. A vacuum pump producing a constant pressure of minus 650 mm Hg causes the solvent molecules to be filtered out through the pores of the membrane filter in a uniform manner. This ultimately results in a well-ordered and layered structure of GO that is left behind. The GO-laden filter paper (Fig. 1c) was then baked on a hot plate at 50 °C for 30 min which allowed for the separation and smooth peeling of free-standing GO papers from the filter membrane. The obtained GO papers, having diameter of approximately 40 mm, were then cut into six equal portions for subsequent thermal reduction studies and for the various structural and electrical characterizations. The final GO papers were found to be approximately 5 µm thick from cross-sectional SEM images and were compact, mechanically stable and flexible. Given that these qualities are increasingly preferred for achieving improved performance in energy storage devices and electrode materials, a facile, low-cost and environmental-friendly approach presented here is deemed an ideal candidate.

To carry out the thermal reduction process, the GO papers were laid flat on a high purity alumina plate and placed at the center of the heating zone of a tube furnace (Fig. 1d). The thermal annealing profile used in this work is as follows. After the loading of the sample, the tube is purged with high purity (99.995%) Ar gas for 15 min at a gas flow rate of 1000 sccm. The flow rate is subsequently lowered to and maintained at 500 sccm. The temperature is then ramped to the set value with a constant rate of 10 °C/min. Once the set temperature is reached, it is allowed to soak for a period of 10 min. The annealing was carried out at various set temperatures ranging from 100 to 600 °C. The heat is then cut off and the sample is allowed to cool down naturally to room temperature before it is unloaded from the furnace. The samples (Fig. 1e) are then stored in a humidity-controlled dry box.

The FTIR spectra were recorded in the wavenumber range of 500–4000 cm⁻¹ with a Bruker Senterra system. Dispersive Raman analysis was conducted using DXR Smart (Thermo Scientific) at a laser excitation wavelength of 532 nm. The lattice parameters were studied using Rigaku Miniflex X-ray diffractometer under a Cu \({K}_{\alpha }\) radiation of wavelength 1.54 \(\AA\) and a 2θ range of 10–80° and a step size of 0.02°. JSM-5800 LV (JEOL) Field-Emission Scanning Electron Microscope (FESEM) was used to track the changes in surface morphology, film thickness and to obtain the elemental analysis via EDX. Surface roughness was measured using AFM imaging using an AFM workshop TT-2 AFM system. The AFM images were analyzed using the free Gwyddion software package. Electrical resistance measurements were performed using an Everbeing four-point probe measurement system in combination with Keysight sourcemeter (B2901A) and a 6.5 digital Keysight multimeter (34461A).

FTIR analysis is a popular technique to determine the presence of functional groups in materials. Figure 2 shows the FTIR traces obtained for the starting GO paper and the GO papers subject to thermal reduction at several temperatures ranging between 100 and 600 °C. It can be clearly seen that for the original GO paper there are four strong peaks in the transmittance data, which confirms the presence of different oxygenated functional groups. The broad peak seen centered around 3300 cm⁻¹ corresponds to that of the hydroxyl (–OH) group, which clearly flattens at 400 °C. This signifies the complete elimination of the hydroxyl groups from the basal plane of GO at this processing temperature. The second important peak that occurs in the FTIR spectrum is that of the carbonyl (C=O) group, seen at a frequency of roughly 1750 cm⁻¹. This peak is also strongly dampened at 400 °C, but it takes 600 °C to fully eliminate the functional group. The elimination happens via the expulsion of gases such as CO₂ and CO from the basal planes and are therefore expected to leave carbon atom vacancies (or defects) behind. This is one of the reasons why rGO can seldom match the superior qualities of pristine graphene. On a positive note, such defects have been reported to be beneficial for certain applications such as sensors and energy storage [79, 80]. The presence of the carboxylic group (–COOH) is confirmed by the third peak at 1400 cm⁻¹, which also gradually disappears as the specimen temperature is increased. The final peak of importance is seen at about 1100 cm⁻¹ and corresponds to the epoxy functional group. Unlike the hydroxyl, carboxy or carbonyl groups which shows signs of peak intensity reduction at lower temperatures, the peak corresponding to the epoxy group is prominent and shows initial signs of weakening only after about 400 °C. This is consistent with earlier reports suggesting that the epoxy groups require higher thermal energy to be dissociated from the lattice [23, 24, 81, 82]. Overall, the FTIR results strongly suggest the presence of oxygenated functional groups in the synthesized GO paper, in agreement with the theoretical models of Lerf [14] and Klinowsky [83]. The systematic weakening and eventual disappearance of the major transmittance peaks with increasing processing temperature confirms the elimination of these functional groups from the GO structure.

Structural changes in GO and rGO can be understood by studying the changes in the characteristic D and G peaks obtained from Raman spectroscopy. The D peak is located at 1350 cm⁻¹ and is a signature of aliphatic carbons forming disordered sp³ regions due to the decoration of graphitic lattice by various oxygen groups and lattice defects. The G peak occurs at 1590 cm⁻¹ and corresponds to the bond stretching of sp² carbons. As evident from Fig. 3a, for a sample of high-quality graphite, the intensity of the G peak is high due to the pristine sp² arrangement. This is indicative of minimal defects in the honeycomb lattice. In striking contrast, the G peak intensity for GO paper is strongly suppressed, indicative of the absence of sp² hybridizations [13]. This is an expected result since it is well known that the synthesis of GO disrupts the initial crystalline structure as a result of the introduction of strong oxidative agents [14, 83].

While a quick visual inspection of the data in Fig. 3a might tempt one to conclude that G and D peak intensities are identical, a qualitative analysis will help shed better light into this matter. This is typically carried out by measurement of the I_D/I_G ratio, an indicator of the degree of defects in the graphitic lattice. The I_D and I_G values obtained for graphite, GO paper and all thermally treated (100–600 °C) GO papers along with the corresponding I_D/I_G ratio values are listed in Table 1. As shown in Fig. 3b, the I_D/I_G ratio is seen to fluctuate around a base value of about 0.92 for all reduction temperatures ranging from 0 to 500 °C. When the temperature is further raised to 600 °C, there appears to be a sizeable rise in the I_D/I_G ratio. This rather interesting behavior can be explained as follows. There are two factors that contribute toward the average defects in the GO material. One is the lowering of defects as the oxygenated functional groups are gradually removed from between the graphitic layers. As evident from the FTIR results of Fig. 2, this removal process begins at about 200 °C. The second factor is, a consequence of the first, wherein atoms of carbon are eliminated from the lattice as the carbonyl and carboxyl groups exit the material in the form of CO₂ and CO gases. The removal of atoms in this manner leaves behind crystal vacancies which now tend to increase the average defects. Therefore, at temperatures between 200 and 500 °C, there is no definitive trend in the I_D/I_G ratios due to the competition between these two mechanisms. At 600 °C, however, there is a clear indication that the defects caused by vacancies takes over and this reflects as a sizeable increase in the I_D/I_G ratio.

It must be emphasized however, that despite the variation in the I_D/I_G being subtle and any meaningful change occurring only beyond a certain threshold temperature, this does not in any way imply that Raman characterization is not effective in studying the impact of thermal reduction. In fact, it is well known that Raman spectroscopy is one of the most effective methods to characterize graphitic materials [79, 84‐86]. Similar difficulties have been previously reported [27, 73], but it must be noted however, that the maximum reduction temperature used in their work was 450 °C, significantly lower than those explored here. Although it would have been very interesting to explore the I_D/I_G trend beyond 600 °C, this was not possible as the high temperatures would cause the GO paper to become brittle, prohibiting any further meaningful assessment. Perhaps, these regimes can be explored by increasing the thickness of the GO paper, or by employing a different reduction strategy [42, 77] altogether. In the latter approach, the obtained GO papers are sandwiched between the confined space of two Si wafers prior to carrying out the reduction at a high temperature of 800 °C. The confinement was reported to have more effectively preserved the physical structure of the GO papers compared to those that were annealed freely. Nevertheless, with the methodology reported in this work, a reasonable expectation would be that, as the reduction temperatures increases further, the vacancy-type defect density may increase significantly causing a larger rise in the I_D/I_G ratio. Unless of course, the defective lattice could be repaired at very high temperatures, in which case the I_D/I_G ratio would decrease. This has been demonstrated [47, 87‐90] but the restoration requires the assistance of a C-source.

Also shown in Table 1 are the L_D (distance between lattice defects) values calculated by two different approaches. According to Lucchese et al. [91], as long as L_D is greater than 6 nm, it can be estimated from the relationwhich also in agreement with the Tuinstra–Koenig model [92]. A more sophisticated method involving Raman mapping was employed by Grimm et al. [93] to account for the inhomogeneous nature of rGO. Here, L_D is estimated from the relationwhere E_L is the laser energy used. From Fig. 3c, it can be seen that at the processing temperature of 600 °C there is a significant drop in the L_D value calculated from both methods. This further confirms the defective nature of the rGO paper.

XRD is an effective tool for revealing the interplanar spacing (d-spacing) using Bragg’s law and the crystalline phases of the material being studied. Figure 4a shows the XRD results obtained for the GO papers reduced at the various reduction temperatures. The two characteristics peaks of interest here, for comparison sakes, are those of the initial GO paper (2-theta = 11.5 °) and a pure sample of graphite (2-theta = 26.55 °). These corresponds to the < 001 > and < 002 > crystalline phases of GO and graphite respectively. As the GO papers are thermally reduced, a gradual shift in the dominant crystalline phase is expected as the oxygenated functional groups are systematically eliminated from the GO framework.

From the results shown in the various panels Fig. 4a, a clear shift from 10.32 ° to 24.8 ° is observed in the major peak of the XRD spectrum as the thermal reduction temperature is increased. At a temperature as low as 200 °C, the prominent < 001 > phase of GO is suppressed by a lot and the < 002 > peak is seen to slowly emerge. For a reduction temperature of 300 °C, the < 001 > peak is completely flattened and a broad < 002 > begins to develop. This suggests that the onset temperature for effective thermal reduction must lie in the 200–300 °C range, in good agreement with the results from the Raman analysis of Fig. 2.

As the temperature is increased further to 500 °C, the < 002 > peak becomes sharpest in intensity. This signifies the improvement of crystallinity as the functional groups are eliminated from the GO lattice. Interestingly, at a reduction temperature of 600 °C, the peak once again broadens, suggesting a loss of crystallinity. This behavior is directly corroborated to the conclusions drawn previously from the I_D/I_G analysis of Fig. 3.

The d-spacing and the full-width half-maximum (FWHM) values can be calculated from the information of the major peak’s positions and widths. The data presented in Table 2 and Fig. 4b reveal that the value of the d-spacing reduces significantly, from 7.6 \(\AA\) for the starting GO paper to 3.5 \(\AA\) for the GO paper thermally reduced at 600 °C. This trend is attributed to the gradual removal to interlayer species, namely the oxygen functional groups [91, 92, 94]. For comparison, the d-spacing value for pure graphite is 3.36 A which agrees with the literature [95]. The quantitate analysis of the FWHM values is shown in Fig. 4c. The FWHM is seen to increase initially as the < 001 > peak broadens and diminishes in intensity. Once the transition from < 001 > to < 002 > phase is underway, the FWHM values show a U-shape trend, which is reflective of an initial improvement in crystallinity followed by a degradation.

The crystal stack height (L_C) values calculated using Scherrer’s equation [96]and the number of graphene layers (N_C = L_C /d) are also reported in Table 2. The variation of L_C with reduction temperature (Fig. 4d) shows a systematic lowering as the temperature is increased. The lowering of L_C has been attributed to exfoliation of carbon layers caused by the thermal annealing [27, 73].

Morphological changes in the reduction process of GO paper were tracked with the help of FESEM micrographs. The various panels of Fig. 5a capture the evolution of the surface morphology for the GO papers reduced at various reduction temperatures. The original free-standing GO paper appears to be dominated by large wrinkles. This is possibly due to the formation of GO agglomerates during vacuum filtration [44] or from turbulent stacking [77]. The surface morphology of the rGO papers, on the contrary, shows a completely different result. For instance, for reduction temperatures as low as 200 °C, the large wrinkles are seen to be significantly much lesser. With further increase of the reduction temperature to 300 °C and 400 °C, the wrinkled structures seem to almost disappear. This is because, as more and more functional groups are eliminated from the GO surface, the effects of turbulent stacking (or agglomeration) are reduced, giving the films a much flatter appearance. Despite this flatter appearance however, the SEM micrographs do not reveal a microscopically smooth surface but instead a surface characterized by a mixture of suppressed wrinkles as well as an increasing level of emerging granularity. This apparent microscopic roughness can be seen in the form of sharp edges and can be attributed to the release of functional groups from the surface of the GO paper [46]. As discussed previously in the Raman analysis, the evolution toward the cluttered surface morphology seen here is once again indicative of competing defect removal mechanisms. At the highest temperatures of 500–600 °C, the SEM micrographs clearly shows a more developed surface, solely dominated by the sharp-edge features. This is strongly suggestive of loss of crystallinity [94] resulting from the propensity of vacancy defect formation caused by CO₂ and CO gases escaping from out of the plane of the GO paper. Since our GO papers are reduced without any physical confinement methods, such as the method reported in Ref [77], the escape of these gases cannot be limited to the in-plane position alone. It is therefore also important to evaluate different strategies for optimizing the reduction efficiency of GO papers and/or obtaining smoother surfaces.

Figure 5b shows the elemental analysis performed on the GO and rGO paper samples. Here, the oxygen and carbon content (wt. %) is plotted as a function of the reduction temperature. The C and O₂ levels remain flat for temperatures up to 200 °C and then strongly change as the temperature is raised further. By 600 °C, the carbon percentage has increased to 85% from an initial value of 60% while the oxygen content has proportionately decreased. At 600 °C, the O₂ level has dropped to a significantly low value of 13.31%. This value corresponds to the remaining functional groups that are not yet expelled from the GO paper. These could very well be the epoxy groups that are known to be energy-costly for removal [23, 24, 94, 97]. Moreover, further carbon wt. % enhancement is not expected due to the propensity of vacancy defects, as noted earlier.

The cross-sectional SEM pictures for the original GO paper and those reduced at temperatures of 200 °C, 300 °C and 600 °C are shown in the panels of Fig. 5c. It has been known from previous works [46, 73, 74] that the thermal reduction of GO results in exfoliation and layer separation due to the dissociation of the functional groups from the basal planes. This can be understood by comparing the micrographs of the original GO paper, in which no clear layer separation can be seen, and the GO paper thermally reduced at 300 °C where the individual graphene sheets can be discerned. For the GO papers reduced at 600 °C, the separation between layers is seen to increase implying volumetric expansion [73] and more effective removal of the functionalities. From an application standpoint, the layer exfoliation and the resulting stacking is a desired quality for electrode materials due to the high surface area than can be accessed for ion and electron diffusion. Overall, these observations provide important clues to understanding the morphological changes occurring at critical temperatures involved in the reduction process.

To shed additional insight on the morphological changes, we have performed AFM scans of the surface of the original GO paper (Fig. 6a) and compared it to the GO paper treated at 400 °C (Fig. 6b). A clear reduction in surface roughness (RMS) is noticed after thermal reduction. The RMS value drops from 0.73 to 0.44 mV, which corresponds to a roughness change of approximately 40%. The resulting smoother surface of the reduced GO sample is attributed to the removal of oxygen functional groups and is consistent with the lowering of wrinkled features seen in the SEM scans. Similar roughness reduction has been reported in the literature [90, 98] albeit with much lower vales than reported here.

In this section, we describe detailed electrical measurements performed using a four-point probe measurement setup. Figure 7a shows the resistance measurements obtained from four batches of vacuum-filtered GO paper samples subject to thermal reduction at temperatures ranging between 100 and 500 °C. The size of the box plot reflects sample to sample variation. The resistance of the GO papers is found to drop very strongly in the temperature range of 200–300 °C. This signifies a transition regime where the GO turns from strongly insulating to strongly conducting. The resistance drops by several orders of magnitude in this temperature window. For higher reduction temperatures, a saturation tendency is observed in the resistance and may point to the fact that further improvement in electrical conduction may be limited. This bodes well with the earlier discussions that have indicated the possibility of a defective GO structure resulting from removal of carbon atoms. Measurements on GO papers reduced at 600 °C have been avoided due to their semi-brittle nature causing difficulty in taking reliable readings.

Figure 7b, shows the low-bias current–voltage curves for the different samples studied. Clearly, for temperatures of 300 °C, 400 °C and 500 °C, classic ohmic behavior is obtained as expected for good conductors. On the other hand, the nonlinear curves obtained at 100 °C and 200 °C, prove that the sample is strongly insulating with very minimal current measured even as the applied voltage is increased by several orders of magnitude. The strong insulating behavior is attributed to the presence of nanometer-scale graphitic domains separated by heavily oxidized regions. The oxidized regions shrink in size as the functional groups are eliminated, resulting in enhanced electrical current [14, 17, 83]. It is further known from temperature-dependent conductivity studies [99, 100], that the electronic transport in rGO is dominated by variable-range hopping (VRH) and thermal activation (TA) pathways in the bias and temperature regimes explored here.

To further analyze the crossover temperature, a different experiment was performed that involved longer reduction times of 30 and 60 min and the inclusion of GO paper reduced at an intermediate temperature of 250 °C. Clearly, as seen in Fig. 7c, at 200 °C the resistance change is minimal even as the reduction time was increased to 30 min. On the other hand, at 250 °C, the resistance has dropped by several order of magnitudes constituting a resistance change greater than 99.9%. This implies strong functional group and crystallographic changes already occurring in the GO paper at this temperature. Therefore, from the observations made thus far, we conclude that the transition temperature must lie in between 200 °C and 250 °C.

To clarify the impact of reduction time on the resistance, we have performed additional measurements as a function of reduction time at 100 °C and 400 °C. While, an increase of the reduction time shows no systematic variation at 100 °C (Fig. 7d), at 400 °C, the resistance is seen to systematically drop with reduction time (Fig. 7e). This is an expected result that corresponds to the gradual removal of functional groups with time, as long as the process temperature is high enough to cause reduction.

Overall, these observations correlate very well with the results of the XRD, FTIR and the elemental analysis discussed in the earlier sections.