3.1. Oxide concentrations
The oxide concentration (OC) was quantitatively determined by Eq. (1), deduced from the method presented by Carvalho et al., in56, based on the peak-to-peak binding energy difference of the high-resolution XPS C1s spectrum.
\(OC={\left(\frac{\varDelta {E}_{C1S}-{E}_{0}}{A}\right)}^{\frac{1}{2}}+{C}_{l}\). | (1) |
In Eq. (1), OC is oxide concentration; \(\varDelta {E}_{C1S}\) is the binding energy (BE) shift between the C sp2 peak and functional group C─OH (hydroxyl) peak; \({E}_{0}\) is the fundamental state of energy or energies imposed by the presence of carbon atoms; \(A\) is a parameter obtained from the fit and related to the energy value of \(52.3 eV\), it can be associated to a slope56; and \({C}_{l}\) is independent of the energies and depends on a concentration due to the influence of the nuclei with a value of \(0.122\)56.
Figure 1b shows OC as a function of TCA; as the synthesis temperature decreases, the OC increases from 0.21 (TCA = 1273 K) to 0.28 (TCA = 773 K). The fit corresponds to a linear relationship given by Eq. (2) and this behavior of OC (TCA) can be attributed to oxides and organic compounds desorption. The negative slope probably indicates that the higher the synthesis temperature is, the more it promotes oxide and organic compounds desorption and, therefore, an OC decrease, as given by:
\(OC=-1.26\times {10}^{-4}*{T}_{CA}+0.37.\) | (2) |
Here, the values of \(-1.26\pm 0.01 \times {10}^{-4} {K}^{-1}\) and \(0.37\pm 0.01\)were related to the slope of Fig. 1b and the extrapolated value of OC at ideal TCA = \(0 K\); respectively, these parameters were obtained from the fitting of the experimental data, as presented in Fig. 1b by employing Eq. (2).
3.2. Compositional and morphological analyses
Figure 2a presents the normalized XPS spectra of 11 GOF samples synthesized to TCA from 773 to 1273 K; peaks are observed at \(\approx 538, 284 eV\)associated with O1s and C1s respectively, evidencing the presence of carbon and oxygen atoms. Given that the general XPS spectra show the majority presence of carbon, as expected, high-resolution C1s and O1s spectra were performed for each of the 11 GOF samples. Figures 2b and 2c show the spectra corresponding to samples S-1173, with their respective deconvolution; the spectra were fitted by the Voigt function (GLP30). The C1s spectra show three bands in the BE range from 270 to 300 eV, these bands were associated with C sp2 (\(284. eV\)) and C sp3 (\(285.2 eV\)) hybridization and functional group of C─OH (\(286.4 eV\))39.
The O1s spectra generally show the presence of four bands in the range from 520 to 540 eV; these bands were associated with functional groups of C═O (\(531.9\pm 0.1 eV\)), C─OH (\(532.9\pm 0.2 eV\) ), C─O─C (\(533.1\pm 0.4 eV\)), and C─O (\(534.2\pm 0.1 eV\)), which supports the same interpretation and values reported by L. Stobinski et al., in39.
Figure 3 presents the SEM micrograph of GOF samples S-973; it is possible to observe fibers with porous tubular and rough surface, lengths in the order of \(3 mm\) and diameters of \(600 \mu m\), surface corrugations with a size of \(40 \mu m,\) approximately, as presented in Fig. 3a. The cross-section of the microchannels was observed with sizes varying from \(1\)to \(20 \mu m,\) as shown in Fig. 3b, and Fig. 3c shows the surface roughness. The microchannels exhibited porous structures with sizes from \(5\) to \(30 \mu m\), as presented in Fig. 3d. This characteristic morphology was observed in all GOF samples studied herein and is typical behavior of GO obtained from RH, as described by Ahiduzzaman M. et al., in57.
3.3. Vibrational properties
The Raman spectra of the GOF samples obtained by varying the TCA from 773 to 1273 K are shown in Fig. 4. Fitting and deconvolution of the Raman spectra served to identify the main vibrational contributions. The D-band ranged from \(1329\) to \(1350 {cm}^{-1}\) for samples with lowest OC (S-1273) and highest OC (S-773), respectively. The G-band varied from \(1600\) to \(1589 {cm}^{-1}\), for lowest OC and highest OC, respectively, as reported by T. Liou and P. Wang in10. The state of the D-band is caused by the presence of defects (disorders, vacancies, and functional groups) and, according to M.S. Ismail et al., in33, its height depends on the number of the sp3 carbon atoms of graphene surface and the number of the defects of the graphene42, while the G-band is caused by the formation of the stretching vibration sp2 carbon atoms and represents the graphitized carbon42. These bands are characteristic of the Raman spectrum of GO, corresponding to the symmetry A1g and the vibrational mode of E2g, respectively58, and overtone bands at high Raman shift of \(2622 {cm}^{-1}\) (2D band), \(2875 {cm}^{-1}\) (D + G band), and \(3100 {cm}^{-1}\) (2D’ band), according to data reported in25. Broadening of these bands is related with the stacking effect of GO monolayers with edges, defects, and sp2 regions25.
3.4. Electrical properties
Figure 5a shows the electrical conductivity variation of GOF samples as a function of the OC. It was found that it decreases OC from 0.25 to 0.21 and increases electrical conductivity from \(4.66\times {10}^{-2}\) to \(4.45 S {m}^{-1}\), as expected; this behavior, \(\sigma =f\left(OC\right)\), is similar to that reported for graphene oxide obtained through other synthesis methods41,45,53,59. This increase can be attributed to the desorption of oxides and organic compounds via thermal decomposition, as a consequence of TCA, which modifies the OC, as reported60 and in graphene oxide nanofibers59. The experimental data of GOF electrical conductivity was fitted employing a polynomial function, as presented by the blue curve in Fig. 5a, as expected and to describe the \(\sigma =f\left(OC\right)\) relation in a semiconductor material, as reported by Van Vechten61. Consequently, it was found that electrical conductivity as a function of OC in GOF samples is given by Eq. (3).
\(\sigma \left(x\right)=d+fx+g{x}^{2}.\) | (3) |
Here, \(x\) is the independent variable associated with OC; \(d\) corresponds to electrical conductivity independent of the OC, with a value of \(161.9\pm 0.1 S/m\); \(f\) was related with the linear factor of the OC that corresponds to a value of \(-1268.4\pm 0.1S/m\); and \(g\) is associated with the nonlinear factor of the OC with a value of \(2483.5\pm 0.1 S/m\). The best fit was obtained with \({R}^{2}=0.97968\), as proposed here.
Equation (4) was used to calculate the \({E}_{g}\) of the GOF samples, described by23.
\(\sigma ={\sigma }_{0}{K}_{B}T\text{*}exp\left(\frac{-{E}_{g}}{2{K}_{B}T}\right).\) | (4) |
Where \({E}_{g}\) is the band-gap energy, \(\sigma\) is the electrical conductivity of the GOF samples presented in Fig. 5a, \({\sigma }_{0}\) is the electronic conductivity independent of temperature, \({K}_{B}\) is the Boltzmann constant, and \(T\) is the temperature.
As seen in Fig. 5b, the \({E}_{g}\) of the GOF samples varies as a function of OC. Band-gap energy shows a variation from \(0.24 eV\) to \(0.48 eV\) by increasing the OC from 0.21 to 0.25, presenting similar behavior to that reported by theoretical studies, which have predicted that increased \({E}_{g}\) is related with increased oxidation, as reported in49,60,62−64. The experimental data of \({E}_{g}\left(x\right)\) was fitted by employing a quadratic function. As illustrated by the red curve of Fig. 5b, it was found that the Van Vechten model describes the experimental results effectively, and it revealed that GOF samples exhibit semiconductor behavior, as expected and given by61.
\({E}_{g}\left(x\right)=a+bx+c{x}^{2}.\) | (5) |
Here, \(x\)is the independent variable associated with OC; \(a\) corresponds to the \({E}_{g}\) independent of the OC with a value of \(8.1\pm 0.1 eV\); \(b\) was related with the linear factor of the OC that corresponds to a value of \(-73.1\pm 0.1 eV\); and \(c\) is associated with the nonlinear factor of OC with a value of \(170.8\pm 0.1 eV\). The best fit was obtained with \({R}^{2}=0.96052\). It is evident that OC modifies the electrical properties of GOFs, as expected60,63,65−68.
In order to elucidate the possible mechanism responsible for semiconductor behavior in GOF samples, measured at room temperature, a correlation was conducted among the influence of OC with \({E}_{g}\) and hydroxyl-epoxy ratio, as presented in Fig. 6. Previous studies reported that oxides in graphene are mainly hydroxyl and epoxy, as functional groups63,65; the presence of these oxides increase the interplanar distance54,63,69. The hydroxyl/epoxy ratio was estimated by employing the area comparison method of XPS spectra as \(\left[1-\left(\frac{{A}_{-O-}}{{A}_{OH}}\right)\right]\) for each GOF sample, as reported70. Figure 6 shows an \({E}_{g}\left(x\right)\) scale with hydroxyl/epoxy ratio, as a function of OC. Consequently, we believe that the presence of multifunctional oxides increases the \({E}_{g}\), as expected for a semiconductor material64, and these changes in the electrical properties of GOF samples can be attributed to the formation at atomic scale of hydroxyl bridges promoted by experimental TCA. The presence of hydroxyl bridges in GOF samples at an atomic scale rearranges the graphene structure and out-of-plane carbon atoms, reducing the average interatomic distance opening the \({E}_{g}\), as reported on graphene oxide obtained from bamboo70. The experimental data of the hydroxyl/epoxy \(\left(\frac{H}{E}\right)\) ratio was fitted by employing a polynomial function, as described by the green line in Fig. 6, given by the equation:
$$\left(\frac{H}{E}\right)\left(x\right)=m+nx+p{x}^{2}.$$
(6)
Here, \(x\)is the independent variable associated with OC; \(m\) corresponds to the hydroxyl/epoxy ratio independent of the OC with a value of \(12.9\pm 0.1\); \(n\) was related with the linear factor of the OC that corresponds to a value of \(-121.9\pm 0.1\); and \(p\) is associated with a nonlinear factor of the OC with a value of \(289.5 \pm 0.1\). The best fit was obtained with \({R}^{2}=0.99641\), as proposed here.
The correlation between \(\left({E}_{g}\right)\) and hydroxyl/epoxy ratio, \(\left(\frac{H}{E}\right)\), shows that decreased \({E}_{g}\) from \(0.48 eV\) to \(0.24 eV\), increases \(\frac{H}{E}\) from \(0.11\) to \(0.58\) (Fig. 6b). The experimental data of \(\frac{H}{E}\) was fitted by using a linear function, as described by the red line in Fig. 6, given by the equation:
$$\left(\frac{H}{E}\right)\left(x\right)=q+rx.$$
(7)
Here, \(x\)is the independent variable associated with \({E}_{g}\); \(q\) corresponds to the \(\frac{H}{E}\) independent of the \({E}_{g}\) with a value of \(-0.36\pm 0.01\); \(r\) was related with the linear factor of the \(\frac{H}{E}\) that corresponds to a value of \(1.92\pm 0.01\). The best fit was obtained with \({R}^{2}=0.97257\), as proposed here.
3.5. Correlation among oxide concentrations, vibrational, and electrical properties
The Raman crystal size and boundary defect density were calculated by employing the equations (8) and (9), respectively, for each GOF sample at different TCA, given by23.
$${L}_{A}\left(nm\right)=4.4\left(\frac{{I}_{G}}{{I}_{D}}\right);$$
(8)
$${{n}_{D}}^{2}\left({cm}^{-2}\right)=107.57\times {10}^{-9}\left(\frac{{I}_{D}}{{I}_{G}}\right).$$
(9)
Here, \({I}_{D}\) is the normalized intensity of the D-band and \({I}_{G}\) is the normalized intensity of the G-band. The increased \(\frac{{I}_{D}}{{I}_{G}}\) ratio in equations (8) and (9) represents the transformation of a disordered structure into an ordered one, as expected. The correlation between OC and vibrational properties shows that a decreased OC increases the boundary defects density from \(3.12\) to \(3.67\times {10}^{-4} {cm}^{-2}\), as presented in Fig. 7a, and decreases crystal size from \(4.88\) to \(3.52 nm\), as shown in Fig. 7b. This can possibly be explained by the desorption of multifunctional oxides and some organic compounds due to thermal decomposition methods employed to synthesize GOF samples, as expected71. The experimental data the defects density was fitted by employing a linear function, as described by the blue line in Fig. 7a, given by the Eq. (10).
$${n}_{D}\left({cm}^{-2}\right)\left(x\right)=\alpha +\gamma x.$$
(10)
Where, \(x\)is the independent variable associated with OC; \(\alpha\) corresponds to the independent term of the OC with a value of \(5.62\pm 0.01 {cm}^{-2}\); \(\gamma\) was related with the linear factor of the OC that corresponds to a value of \(-9.13\pm 0.01 {cm}^{-2}\). The best fit was obtained with \({R}^{2}=0.99361\), as proposed here. The experimental data the crystal size was fitted by employing a linear function, as described by the red line in Fig. 7b and given by:
$${L}_{A}\left(nm\right)\left(x\right)=\delta +\epsilon x.$$
(11)
Here, \(x\)is the independent variable associated with OC; \(\delta\) corresponds to the independent parameter of the OC with a value of \(-1.33\pm 0.1 nm\); \(\epsilon\) was related with the linear factor of the OC that corresponds to a value of \(22.47\pm 0.1 nm\). The best fit was obtained with \({R}^{2}=0.99212\), as proposed here. The correlation between oxide concentration and vibrational properties can be described by expressions (10) and (11), as proposed here.
The correlation between electrical conductivity and vibrational properties, like defects density and crystal size, shows that increased electrical conductivity from \(4.66\times {10}^{-2}\) to \(4.45 S/m\) increases defects density from \(3.19\times {10}^{-4}\) to \(3.58\times {10}^{-4} {cm}^{-2}\) and decreases crystal size from \(4.65\) to \(3.64 nm\), as presented in Figs. 7c and 7d. Thus, it can be deduced that low OC increases the characteristic relaxation time of the electric charge carrier dispersion process, while high OC decreases this characteristic time, due to a higher presence of impurities, mainly hydroxyl groups. These behaviors are very important, demonstrating that electrical and vibrational properties are tuned by OC; also, revealing the multifunctional effect of hydroxyl and epoxy groups present in GOF samples.
The experimental data the defects density was fitted by employing a polynomial function, as described by the red line in Fig. 7c and given by the Eq. (12).
$${n}_{D}\left({cm}^{-2}\right)\left(\sigma \right)=\xi +\phi \sigma +\nu {\sigma }^{2}.$$
(12)
Where, \(\sigma\)is the independent variable associated with electrical conductivity; \(\xi\) corresponds to the independent term of the electrical conductivity with a value of \(3.35\pm 0.01 {cm}^{-2}\); \(\phi\) was related with the linear factor of the electrical conductivity, that corresponds to a value of \(0.12\pm 0.01\frac{ m}{S}{cm}^{-2}\); and \(\nu\) is associated with the nonlinear factor of the electrical conductivity with a value of \(-0.01\pm 0.01 \frac{{m}^{2}}{{S}^{2}}{cm}^{-2}\). The best fit was obtained with \({R}^{2}=0.98047\), as proposed here.
The experimental data the crystal size was fitted by employing a polynomial function, as described by the blue line in Fig. 7d and given by:
$${L}_{A}\left(nm\right)\left(\sigma \right)=\zeta +\rho \sigma +ϵ{\sigma }^{2}.$$
(13)
Here, \(\sigma\)is the independent variable associated with electrical conductivity; \(\zeta\) corresponds to the independent term of the electrical conductivity with a value of \(4.21\pm 0.01 nm\); \(\rho\) was related with the linear factor of the electrical conductivity that corresponds to a value of \(-0.29\pm 0.01 \frac{m}{S}nm\); and \(ϵ\) is associated with the nonlinear factor of the electrical conductivity with a value of \(0.03\pm 0.01 \frac{{m}^{2}}{{S}^{2}}\text{n}\text{m}\). The best fit was obtained with \({R}^{2}=0.98079\), as proposed here. The correlation between electrical conductivity and vibrational properties can be described by expressions (12) and (13), as proposed here.