Experimental evaluation of eco-friendly exfoliation strategies for Tour-method graphene oxide

Abstract

The exfoliation of graphite oxide (GrO) represents a critical stage prior to processes such as the reduction or functionalisation of graphene oxide (GO), directly affecting its final properties. In particular, GrO synthesised by the Tour method, which offers advantages over conventional methods such as the Hummers method (higher safety, better oxidation degree and homogeneity of the material), has been scarcely studied in terms of the control of its specific surface area (SSA), a fundamental parameter for its performance in technological applications. In this work, the optimisation of the exfoliation process of GrO prepared by the Tour method is proposed through the systematic comparison of different eco-friendly techniques: ultrasonic bath, ultrasonic probe, ball milling and glucose-assisted ball milling. A rigorous Design of Experiments (DE) and a comprehensive statistical analysis were carried out, using carbon content, C (%), and SSA as main responses. In addition, complementary characterisations by X-ray diffraction (XRD, for number of stacked layers, "n", crystal size, L, and diffraction peak position, 2θ) were performed to deepen the structural analysis. The results reveal that ball milling without additives achieved the highest SSA (~ 71 m² g⁻¹), followed by glucose-assisted milling (~ 54 m² g⁻¹), while ultrasonic techniques offered significantly lower values (6–30 m² g⁻¹). This study provides a solid and rigorous basis for understanding and optimizing the properties of GO according to the requirements of the specific applications for which it is intended and a study of the differences between ball milling and ultrasonic exfoliation in GO normally applied in graphite exfoliation.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-42185-9.

Introduction

Graphene-based materials (GBMs) are highly versatile, due to both the intrinsic properties of graphene and the additional characteristics that emerge from its final structure. In its pure form, graphene is distinguished by its electrical conductivity, strength, lightness, flexibility, and transparency^1–4. However, when GBMs are synthesised in various forms, such as particles, fibres, papers, layers, sheets, or three-dimensional structures^5–7, they acquire additional properties like anisotropy and porosity^8,9. The ability to form fibres, two-dimensional or three-dimensional structures is particularly significant, as it introduces new properties that are absent in pure graphene, greatly expanding its range of applications. Furthermore, GBMs can be functionalized^10–13, doped with various elements and compounds^14–16, or chemically modified¹⁷ to enhance or add specific properties, increasing their efficiency in targeted applications. This combination of intrinsic and acquired properties allows GBMs to have a broad spectrum of applications, ranging from sensors^18,19, supercapacitors and electronic devices^3,20 to medical applications, such as nanoparticles for controlled drug delivery^21–24.

GBMs are not typically manufactured from pure graphene due to its low stability, the complexity of its production on an industrial scale, and the high production and energy costs involved. Instead, many GBMs syntheses are carried out using a more stable and easily obtainable graphene derivative, typically micron-sized graphene oxide (GO) sheets²⁵. GO is a type of GBM that can be used directly in applications such as sensors^26,27 or therapeutic uses^28,29, or it can be reduced or functionalized for applications like supercapacitors^30–32 or the capture of radioactive elements^33,34. Additionally, more complex structures in the form of fibres^35–38, films^39–41, or aerogels^42–45 can be obtained through various methods, such as hydrothermal processing^46–48, chemical reduction^49–51, or 3D printing⁵², among others. These complex structures are then used in applications like catalysis^53,54 or energy storage^55–57. All these applications are possible because these materials retain many of the outstanding characteristics of graphene⁵⁸.

The use of GO offers significant advantages in the synthetic processes of GBMs. The oxygenated functional groups on GO sheets confer polarity, making GO highly soluble in water, which is often used as a solvent in most synthetic processes of GBMs^6,59. Some syntheses, such as hydrothermal or chemical reduction techniques^60,61, involve the removal of these oxygenated groups from the surface of GO sheets. This removal increases the hydrophobicity of the sheets, allowing them to couple in a polar medium, leading to the formation of three-dimensional structures. These methods also help in the partial recovery of electrical conductivity by partially regenerating the original graphene structure. Additionally, the presence of oxygenated groups provides GO with versatility, allowing it to be reduced to the desired degree^34,62–64 and creating active sites that can be functionalized with different elements and compounds^65,66, depending on the specific requirements of each application.

GO is usually obtained from graphite oxide (GrO), synthesised previously by graphite oxidation using some typical method, such as Brodie⁶⁷, Hummer⁶⁸ or Tour^52,69. Among them, the Tour method offers significant advantages, including a higher oxidation degree, better structural homogeneity, and increased safety during synthesis^70,71. This process is followed by the GrO to GO step, which involves some exfoliation treatment breaking the van der Waals forces between the GrO sheets. The exfoliation process controls the final number of stacked layers ("n") and defect density in GO, which in turn determine its specific surface area (SSA). A higher degree of exfoliation results in fewer layers and larger SSA³⁴, enhancing the material’s active area, reactivity and suitability for various applications, such as sensors, energy storage, or catalysis, where material’s SSA directly impacts its efficiency⁷².

There are, currently, a multitude of methods for GrO exfoliation, ideally aiming to achieve single GO sheets with 100% yield⁷³. However, generally, GOs are obtained from a few stacked sheets and yields between 1 and 95%^74,75. Exfoliation methods can be classified from the simplest, such as ultrasonic (US) bath or probe^73,76,77, to the most sophisticated and complex, such as supercritical fluid exfoliation⁷⁸. Some of them involve exfoliation by applying heat, using an oven, microwave, or other heat sources^34,79–82. At elevated temperatures, graphene oxides lose oxygenated functional groups in the form of CO and CO₂^4,83,84. The bubbles generated during this process separate the sheets, leading to thermal exfoliation and a reduction of the GrO, which results in the partial recovery of electrical conductivity⁸⁵, making them suitable for applications in electronics⁸⁶, battery electrodes^87,88, and supercapacitors^89–92. Nevertheless, thermal exfoliations are more prone to inducing fractures and structural defects, and are less suitable for applications that require subsequent functionalization. Many low temperature exfoliation processes involve using molecules, such as surfactants to intercalate and expand the interplanar distance between layers⁹³. However, these methods have notable drawbacks, including the toxicity of the reagents used and the need for a subsequent purification stage. In contrast, I. Ogino et al. (2015) have demonstrated that GrO exfoliation can be achieved through freeze-thaw cycles of water, a method that avoids toxic reagents⁹⁴, though it presents challenges in terms of scalability.

In this context, mechanical methods, such as ultrasonic and ball milling methods are more accessible and cost-effective, requiring fewer toxic reagents, making them attractive options for the exfoliation of GrO^73,77,95,96. Ultrasonication, in particular, relies on the propagation of ultrasound waves, whether from a probe or bath, propagate through water, creating bubbles that collapse and separate the graphite oxide sheets through shear forces. In the work of Tyurnina et al. (2020), the correlation between the size and behaviour of microbubbles and the power and frequency of ultrasonic waves is demonstrated. The authors showed that at low frequencies, large bubbles are generated, leading to violent collapses. This favours higher yields but also produces more fractured materials with a higher defect density. In contrast, at higher frequencies, smaller bubbles are formed, generating repeated oscillations and gentler collapses, resulting in more controlled exfoliation, resembling a “book-page opening” mechanism. In this case, the materials obtained after exfoliation exhibit fewer defects and thinner sheets, although with lower yields⁹⁷. Additionally, when using an ultrasonic probe that is in direct contact with the GO solution, slight local temperature increases can alter the carbon-to-oxygen ratio (C/O)^95,98. Despite these drawbacks, ultrasound exfoliation remains as an attractive option due to its simplicity, speed, low cost, and eco-friendly nature. On the other hand, ball milling exfoliation^76,99, applies normal and shear forces to separate the layers. This technique has a significant effect on the material structure. During exfoliation, the process often induces partial reduction⁹⁹, leading to the loss of some oxygenated functional groups that are essential for graphene’s chemical reactivity. The shear forces not only separate the layers but also promote oxygen removal, compromising the material oxidised structure. Under more severe conditions, normal impact forces dominate, potentially causing structural damage and fracture of the GO sheets¹⁰⁰. Compared to sonication, ball milling typically results in materials with a higher level of structural defects, which can negatively affect their physical and chemical properties⁷⁶. Nevertheless, ball milling remains an attractive option due to its capacity for large-scale production and its efficiency in producing large graphene sheets. It is evident that comprehensive knowledge of the influence of exfoliation process parameters on the properties of the resulting GO is crucial for its optimization and use in different applications. In Fig. 1 the effects of different exfoliation conditions by US and ball milling on GrO sheets are summarized.

Fig. 1.

Exfoliation mechanisms of GrO by US and ball milling under harsh and mild conditions.

Although exfoliation processes represent a crucial stage in the synthesis of GBMs, there is a lack of systematic studies addressing how the experimental conditions of these processes affect key structural properties of GO, such as the SSA and the number of stacked layers. The Tour method offers a safer and more efficient route for the synthesis of GrO compared to more classical methods, but, the majority of studies reported in the literature on exfoliation processes to obtain GO focus on GrO produced by methods such as Hummer´s^101–103, mostly addressing exfoliation through high-temperature approaches^4,67,104 rather than mechanical processes.

This study addresses that gap by performing a comparative analysis of four eco-friendly mechanical exfoliation strategies: US bath, US probe, ball milling, and glucose-assisted ball milling. Through a rigorous Design of Experiments (DE), the study provides quantitative insights into how these methods affect carbon content, SSA, layer thickness, and crystalline structure, offering practical guidance for tailoring GO properties for specific applications.

This study makes a novel contribution by combining a DE approach for statistical optimisation with comprehensive multi-technique structural characterisation, establishing a benchmark for green, scalable, and application-oriented exfoliation of Tour-method GrO, and further exploiting the advantages of this oxidation method.

Materials and methods

Materials

Graphite powder (< 200 μm, synthetic) was purchased from Sigma Aldrich (Riedstr. Steinheim, Switzerland). Phosphoric acid (H₃PO₄, 85% w/w), concentrated sulfuric acid (H₂SO₄, 98% w/w), potassium permanganate (KMnO₄) and hydrogen peroxide (H₂O₂, 33% v/v) were of analytical grade from Merck (Darmstadt, Germany); hydrochloric acid (HCl, 37% w/w) was from GlobalChem (Cosela S.L., Spain); Dialysis tubing cellulose (flat width 33 mm) was from Sigma Aldrich (USA), and ultrapure water (Resistivity ≥ 18.2 MΩ cm) was from Milli-Q system (Millipore Bedford, Bedford, MA, USA). D (+)-Glucose (Millipore, Darmstadt, Germany).

Methods

Synthesis of graphite oxide based on the Tour method

For the synthesis of GrO, the Tour method, optimised in our previous work²⁵ has been employed. For this purpose, 300 mL of a mixture of H₂SO₄ and H₃PO₄ (9:1 v/v) was prepared to which 1.5 g of graphite was added with stirring. Then, 12 g of KMnO₄ were added slowly, at a time range of 30 min. The mixture was left stirring for 12 h at 60 °C. After this time, 250 mL of distilled water, cooled in ice, were added drop by drop. This process lasted about 30 min. Finally, 150 mL of H₂O₂ were added, generating a light-yellow mixture. The reaction mixture was allowed to settle overnight. The next day the supernatant liquid was removed with a pipette and the solid was washed twice with 150 mL of a 10% HCl solution. The obtained GrO was centrifuged for 1 h at 8000 rpm, diluting it in water up to 350 mL. This process was repeated until the centrifugation liquid reached a pH of 5–6. Then, the product was placed in Petri dishes, frozen at −87 °C, for 48 h and freeze-dried for 5 days (− 111° C, 0.043 hPa).

Exfoliation methods

Four exfoliation methods were used to obtain GO from the previously synthesised GrO: US bath, US probe, and ball mill (with and without glucose) (Fig. 2). For the exfoliation with US bath and probe, 50 mL of aqueous GrO samples of different concentrations were prepared. In case of ball mill exfoliations, dry GrO samples were employed. The different exfoliation methods are briefly described in Table 1.

Fig. 2.

(a) GrO powder; (b) Ultrasonic Bath; (c) Ultrasonic Probe; (d) Ball Mill.

Table 1.

Variables employed in the exfoliation methods.

Method	Variable
	[GrO] (mg mL−1)	Time (min)	Power (%)
Ultrasonic bath	0.5–2.5	30–90	–
Ultrasonic probe	0.5–2.5	30–90	30–100

Method	Variable
	[GrO] (g)	Time (min)	Glucose (g)	Frequency (Hz)
Ball mill	0.025–0.1	5–8	–	7–10
Glucose-assisted ball mill	0.025–0.1	5–8	3.5–5.5	7–10

Ultrasonic Bath: A P SELECTA ULTRASONS ultrasonic bath was used, keeping the water at a temperature of approximately 2°C using ice, to avoid overheating of the samples. The variables studied were sonication time and GrO concentration (Table 1). After carrying out each exfoliation experiment, the aqueous solution of exfoliated GO was frozen at − 86 °C for 48 h and subsequently freeze-dried at − 111 °C and 0.043 hPa for 3 days.

Ultrasonic Probe: A BANDELIN SONOPLUS ultrasonic probe has been used. The samples were placed in an ice container to avoid overheating during the process. The variables studied were time, power and GrO concentration (Table 1). After carrying out each exfoliation experiment, the aqueous solution of exfoliated GO was frozen at − 86 °C for 48 h and subsequently freeze-dried at − 111 °C and 0.043 hPa for 3 days.

Ball mill: A RETSCH MM400 ball mill has been used. The variables studied were time, frequency and the amount of GrO (Table 1). After performing each exfoliation experiment, the obtained samples were directly characterized.

Glucose-assisted ball mill: A RETSCH MM400 ball mill has been used. The variables studied were time, frequency and the amounts of GrO and glucose (Table 1). To carry out the experiments corresponding to this method, the different mixtures of GrO and glucose were prepared in the appropriate proportions. After the exfoliation process, each mixture was dispersed in 50 mL of water in order to dissolve the glucose, which was subsequently removed using reverse osmosis membranes. After removal of glucose, the aqueous solution of exfoliated GO was frozen at − 86 °C for 48 h and subsequently freeze-dried at − 111 °C and 0.043 hPa for 3 days.

Design of Experiments

A screening DE was carried out using STATGRAPHICS Centurion XVII. SSA and the percentage of carbon (C (%)) were chosen as study responses because they are two of the characteristics that most influence the applicability and subsequent efficiency of the synthesised materials. For each exfoliation method, the experimental factors that most influence each process were chosen and for each of them two values were defined. This combination defines a factorial design with two levels for each method. All experimental factors and values used for the four exfoliation methods studied are listed in the Results and discussion section.

Moreover, in order to a complete understanding of each method performance an exhaustive characterisation of each of the material synthesised was performed applying structural evaluation by XRD.

In XRD analysis, the number of stacked layers in the graphite oxide, "n", was obtained by Eq. (1),

1

Where “L” is the crystal size and “d” is the interplanar spacing.

Furthermore, L has been calculated using the Scherrer Eq. (2)²⁵:

2

Where λ is the X-ray wavelength, β is the line broadening at half the maximum and θ is the Bragg angle.

To compare the results obtained from the different experiments made in each optimisation the Student’s T-test was applied. First of all, Fisher’s F-test was applied to compare the standard deviations (s) and to calculate the degrees of freedom (d.f.) of each group of data. With the data obtained, the averages were compared using Student’s T-test¹⁰⁵.

To calculate the F value (F_Cal) Eq. 3 was applied.

3

Where s₁ is the standard deviation of group 1 of results and s₂ is the standard deviation of group 2 of results.

If the value F_Cal is lower than the tabulated F value (F_crit) it can be considered that the standard deviations of the two groups of measurements are equal so it can be applied the Eq. 4 to calculate the degrees of freedom (d.f.) for tabulated values correspond to the T-test (T_Crit). Otherwise, F_Cal> F_Crit it is necessary to applied Eq. 5.

4

5

Where n₁ corresponds to the number of experiments in the condition one and n₂ corresponds to the number of experiments in the condition two.

Once the degrees of freedom have been calculated, it is possible to establish T_Cal by applying the Eq. 6 provided that the hypothesis F_Cal< F_Crit is fulfilled. The values obtained (T_Cal) were compared with the values tabulated (T_Crit) in the tables¹⁰⁵.

6

In case the standard deviation are not comparable, F_Cal> F_Crit, Eq. 7 has to be applied for the calculation of T_Cal.

7

Where is the average value of group 1 of results and is the average value of group 2 of results and s₁ and s₂ are the standard deviation values of group 1 and group 2, respectively.

Characterisation

To evaluate textural parameters, in particular, the interaction potential between the pore surface and gas N₂, adsorption/desorption processes were applied at 77 K temperature, employing ASAP 2020 (Micromeritics, Norcross, GA, USA) equipment. The Brunauer–Emmett–Teller (BET) method was applied to determine the SSA.

The total carbon content was determined by combustion using a TruSpec CHN elemental analyser (LECO, Benton Harbor, MI, USA). This method determines the carbon content by heating to a temperature of at least 900 °C in the presence of oxygen gas. Then, GO samples were oxidised and/or volatilised to carbon dioxide (CO₂). The amount of CO₂ was measured using an infrared detection method.

Structural studies were performed employing X-ray diffraction (XRD) equipment, X’Pert Prodiffractometer (Malvern PANalytical), with Cu K_α radiation (λ = 1.54 Å) with an operating at 45 kV and 40 mA. XRD data were collected in θ–θ configuration in the angular range of 5 < 2θ (°) < 80 with a 0.017 step size. The reflection [001] 2θ (°) of GO was measured, which is typically observed in the range of 10–12 2θ (°), corresponding to an interlayer distance of 7–8 Å. The increasing interlayer spacing of the GO samples suggests that different levels of oxygen containing groups were attached to the graphite lattice.

FTIR analyses were performed using a PerkinElmer Spectrum 100 spectrometer (PerkinElmer, USA). The spectra were recorded in transmittance mode over the wavenumber range of approximately 4000–400 cm⁻¹, with a resolution of 1 cm⁻¹. Baseline correction was subsequently applied using the asymmetric least squares (ALS) method, and the spectra were normalized with respect to their maximum intensity. Scanning electron microscopy (SEM) was employed to assess the morphology and the degree of agglomeration/exfoliation of the GBMs. The observations were carried out using a FEI Inspect™ F50 microscope (FEI Company, USA), operated at an accelerating voltage of 5 kV in secondary electron mode.

Results and discussion

Ultrasonic bath

In this DE, two factors (time and GrO concentration) were proposed with two different values (Table 2). Due to the very small amount of exfoliated material obtained after each run, several repetitions were required and subsequently pooled into a single representative sample per experimental condition and exfoliation method, thus yielding one effective replicate for analysis.

Table 2.

DE for ultrasonic bath exfoliation.

US Bath Exfoliation
Sample	Variable
	Time (min)	GrO (mg mL−1)
GO-B1	90	0.5
GO-B2	30	0.5
GO-B3	30	2
GO-B4	90	2

This Design of Experiments resulted in four experiments (Table 2). All materials obtained from each experiment was characterised by elemental analysis (EA), XRD and by applying BET methodology. The textural, chemistry and structural results obtained in ultrasonic bath optimisation are summarised in Table 3.

Table 3.

The textural, chemical and structural results obtained in ultrasonic bath optimisation.

US Bath	Variable	Time (min)		GrO (mg mL−1)
	Value	30	90	0.5	2
SSA (m2 g−1)	Range	17.0–29.8	7.7–11.9	11.9–17.0	7.7–29.8
	Average	23.4	9.8	14.4	18.7
	SD	9.1	3.0	3.6	15.6
	TCal	2.01		0.38
	TCrit	4.30		4.30
C (%)	Range	39.2–39.9	39.8–40.5	39.2–39.8	39.9–40.5
	Average	39.6	40.1	39.5	40.2
	SD	0.5	0.5	0.4	0.4
	TCal	1.17		1.71
	TCrit	4.30		4.30
2θ (°)	Range	9.8–10.4	10.6–10.6	9.8–10.6	10.4–10.6
	Average	10.1	10.6	10.2	10.5
	SD	0.5	0.06	0.6	0.08
	TCal	1.49		0.65
	TCrit	12.71		4.30
n	Range	6.8–10.2	10.2–12.3	6.8–12.3	10.2–10.2
	Average	8.5	11.2	9.6	10.2
	SD	2.4	1.5	3.9	0.01
	TCal	1.37		0.22
	TCrit	4.30		12.71

Table 3 shows the range, average, standard deviation (SD) and T-Student calculated value (T_Cal) of the results corresponding to SSA and C (%), used for the DE and the position 2θ (°) and number of layers “n”, obtained from the XRD analysis. The complete results can be found in Table AI in Online Appendix A.

From the XRD analysis, the position data (2θ (°)), that correspond to the diffraction maximum of GO [001] reflection, typically located in the range of 9–12°^106–109, and the number of stacked layers “n” have been presented. The average positions of the diffraction maxima are in the range of 10.1–10.6 °2θ (Table 3). These values are very similar indicating reproducibility in the synthesis method of GrO. That fact is corroborated by T-Student test with a T_Cal value, in both cases (time and GrO concentration), lower than T_Crit (12.71 for time and 4.30 for GrO concentration) for a 95% confidence interval.

Values of number of stacked layers “n” show a more pronounced difference in the results obtained at the two sonication times. In this case, 30 min of sonication results in lower average “n” values (8.5) that are in agreement with higher SSA values (average value of 23.4 m² g⁻¹. It is observed that according to the T_Cal values, no value exceeds the T_Crit (4.30 for time and 12.71 for GrO concentration), which indicates that none of the results obtained are statistically significantly affected by variations in time or GrO concentration for a 95% confidence interval.

In the case of the results obtained from C (%) and SSA analysis, a more comprehensive evaluation is shown below thanks to the results obtained from the experimental design. SSA results are in the range of 7.7 to 29.8 m² g⁻¹, similar to other studies¹¹⁰ and C (%) are in the range of 39.2 to 40.5% showing that the different variables used provide a material with a very similar degree of oxidation. From the statistical analysis of the results, it is possible to point out that the highest T_Cal value is observed in the case of the comparison of the SSA resulting at different sonication times (2.01), being the most influential factor in this case, although it does not exceed the critical value (4.30).

The DE employed to evaluate the ultrasonic bath exfoliation method provides information on the variables used as well as the interaction between them. Thus, Fig. 3a and b show the Pareto chart and main effects plot respectively, corresponding to the SSA analysis. According to the Pareto chart (Fig. 3a), it can be seen that time and the interaction between time and GrO concentration have a negative effect on SSA. However, this effect is not statistically significant in agreement with the results in Table 3. Whereas a higher GrO concentration results in an increase in SSA. Generally, lower GrO concentrations lead to more efficient exfoliations. In this case, the DE considers the fixed mean time of 45 min to analyse the influence of GrO concentration. This time may be optimal for higher concentrations but too long for lower concentrations, where excess energy starts to be used to break up individual GO lamellae, which decreases their SSA. Generally, times above 30 min are high for low concentrations, while high concentrations (≥ 0.2 mg mL⁻¹) require times longer than 60 min to result in complete exfoliation without producing appreciable damage to the lamellae.

Fig. 3.

a and b Pareto chart and main effects plot respectively for SSA.

In the main effects plot (Fig. 3b), it is corroborated that experiments carried out at shorter times, 30 min, and higher GrO concentrations generate larger SSA. SSA decreases with increasing sonication time, from approximately 24 m² g⁻¹ to 9 m² g⁻¹. This effect may be explained by a mechanism of shear forces at the early stages of the process, which exfoliate GrO by gently separating its layers; however, at prolonged treatment times, the sheets become thinner and therefore more susceptible to excessive fragmentation⁹⁷.

Figure 4a and b show the Pareto chart and main effects plot, respectively, corresponding to C (%) of the as-prepared samples. In this case, both time and GrO concentration have a positive effect, increasing C (%). On the other hand, the interaction between both variables generates a negative effect, according to the Pareto chart (Fig. 4a), in which we can also see that, none of the effects is statistically significant, according to the results in Table 3. The carbon content increases with increasing sonication time, from approximately 39.5 to 40.1% due to the reduction and removal of some of the oxygenated functional groups as a result of acoustic cavitation that generates high pressure and much localised high temperatures and shock waves, that can physically break bonds⁷⁷. On the other hand, a high GrO concentration can facilitate more effective interactions and redox reactions.

Fig. 4.

a and b Pareto chart and main effects plot respectively for C (%).

Ultrasonic probe

In the ultrasonic probe employed, time and power parameters can be selected, then three factors (power, time and GrO concentration) were proposed with two different values in this DE. The applied conditions are shown in Table 4, resulting in eight experiments.

Table 4.

DE for ultrasonic probe exfoliation.

US probe exfoliation
Sample	Variable
	Time (min)	GrO (mg mL−1)	Power (%)
GO-P1	30	2	30
GO-P2	90	2	100
GO-P3	30	2	100
GO-P4	30	0.5	100
GO-P5	90	0.5	100
GO-P6	90	2	30
GO-P7	90	0.5	30
GO-P8	30	0.5	30

As previously reported in ultrasonic bath study, textural, chemistry and structural characterisation, of each material obtained, was performed. Table 5 shows the results of the range, average, SD and T_Cal for SSA, C (%), diffraction peak 2θ (°) and “n” corresponding to the DE of the ultrasonic probe. The complete results can be found in Table All in Online Appendix A.C

Table 5.

The textural, chemical and structural results obtained in ultrasonic probe optimisation.

US probe	Variable	Power (%)		Time (min)		GrO (mg mL−1)
	Value	30	100	30	90	0.5	2
SSA (m2 g−1)	Range	6.7–10.1	8.5–9.9	6.7–10.1	8.5–9.6	9.6–9.9	8.5–10.1
	Average	8.7	9.2	8.8	9.1	9.0	8.9
	SD	1.5	0.7	1.6	0.6	1.5	0.8
	TCal	0.55		0.40		0.03
	TCrit	2.45		2.45		2.45
C (%)	Range	40.6–43.2	38.4–43.0	38.4–41.5	40.6–43.2	40.1–43.0	38.4–43.2
	Average	41.6	40.8	40.3	42.1	41.3	41.1
	SD	1.1	2.0	1.4	1.2	1.3	2.0
	TCal	0.75		1.93		0.17
	TCrit	2.45		2.45		2.45
2θ (°)	Range	9.8–10.8	10.1–10.8	9.8–10.8	10.0–10.8	10.1–10.8	9.8–10.8
	Average	10.2	10.5	10.4	10.4	10.4	10.3
	SD	0.4	0.3	0.5	0.4	0.4	0.4
	TCal	1.18		0.07		0.51
	TCrit	2.45		2.45		2.45
n	Range	7.1–7.8	7.1–15.6	7.3–15.6	7.1–8.7	7.1–12.4	7.1–15.6
	Average	7.5	11.0	10.8	7.7	8.8	9.7
	SD	0.3	3.8	4.0	0.7	2.4	4.0
	TCal	1.81		1.52		0.39
	TCrit	2.45		3.18		2.45

It is observed that the SSA results present average values between 8.7 and 9.2 m² g⁻¹, being the variation very slight when varying the different parameters. This is corroborated by T_Cal results, which show very low values, between 0.03 and 0.55, and below the critical value (2.45). The average C (%) values are in the range of 40.3–42.1%. According to the T_Cal results, the maximum value is obtained in the case of the comparison of samples obtained at different times, with a T_Cal corresponding to 1.93. Therefore, time is the experimental factor that has the greatest influence on the C (%). However, none of the T_Cal are statistically significant as it does not exceed the critical value (2.45).

According to the data obtained from the peak position of GO [001] reflection, it can be seen that the maximum average value corresponds to samples obtained at a power of 100%, corresponding to 10.5°. On the other hand, the lowest average value, 10.2°, corresponds to 30% power. Although the average values of peak position are very similar, the value of T_Cal is relatively high compared to the other factors, which means a wider dispersion of results. However, none of the T_Cal (time, GrO concentration and power) are statistically significant for the results obtained for the peak position, 2θ (°). These results are slightly higher, i.e. the position of the peak is shifted to the right as was reported earlier by Mellado et al. ⁷⁷.

The results of “n” show average values between 7.5 and 11.0, with lowest values corresponding to samples obtained at 30% power (7.5) and time of 90 min (7.7). To longer times, more exfoliated GO are produced and higher SSA, while lower powers result in higher C (%). As in the rest of the results discussed, none of the T_Cal exceeds the critical value, so the differences of “n” obtained in the different experimental conditions are not statistically significant.

Figure 5a and b show the Pareto chart and main effects plot, respectively, corresponding to the SSA. According to the Pareto chart, it is observed that the most influential experimental factors are the interactions between GrO concentration and power or time, having a negative effect, decreasing the SSA. However, no experimental factor is statistically significant.

Fig. 5.

a and b Pareto chart and main effects plot, respectively, for SSA.

On the other hand, the main effects diagram (Fig. 5a) shows that both sonication time and applied power promote the increase of SSA. Longer sonication times allow a greater interaction of the ultrasonic waves with the material, promoting the dispersion and exfoliation of the graphite oxide. In addition, the increase of the energy, supplied with time, favours the breaking of agglomerations and improves the distribution of graphene oxide in the solution. At higher powers, the ultrasonic waves generate higher cavitation energy, resulting in more efficient exfoliation of GrO to GO. Higher power helps to break the van der Waals forces between the GrO layers, which contributes to increase the SSA. While an increase in GrO concentration has the opposite effect, slightly decreasing the SSA of the GOs obtained. At higher GrO concentrations, the particles tend to agglomerate more, which reduces the efficiency of sonication to disperse and exfoliate them completely, decreasing the SSA. In addition, at higher concentrations, sonication may be less efficient due to the higher viscosity of the solution, which reduces energy transfer⁷³.

These results corroborate the results obtained in Table 5 where higher sonication powers generate higher SSA and structures closer to GO.

Figure 6a and b show the Pareto chart and main effects plot, respectively, corresponding to the analysis obtained for C (%). In the Pareto chart, it can be seen that the experimental factor that has the greatest influence is the sonication time, increasing the carbon content.

Fig. 6.

a and b Pareto chart and main effects plot respectively for C (%).

In the main effects diagram (Fig. 6b), it can be seen that GrO concentration and power have a negative effect on the C (%), while time generates higher C (%). The carbon content increases with sonication time, from approximately 40.4 to 42.1%. As sonication time increases, the ultrasonic waves promote better exfoliation and dispersion of the graphite oxide, allowing a better exposure and possibly better removal of surface oxygen molecules⁷⁷. On the other hand, the C (%) decreases slightly with increasing GrO concentration, from approximately 41.3 to 41.1%. At higher graphite oxide (GrO) concentrations, sonication may be less effective in dispersing and completely exfoliating the particles, which may lead to less removal of surface oxygen molecules producing a slight decrease in carbon content. In addition, a higher viscosity of the solution negatively affects the elimination of oxygen molecules. An increase of the sonication power also contributes to a decrease of the C (%). Higher powers can cause a higher fragmentation of the material and possible re-oxidation of the surfaces, which could explain the observed decrease in carbon content^73,77.

Ball mill

The DE applied in this developed method is shown in Table 6. In the ball mill employed, time and frequency parameters can be selected, then in this Design of Experiments three factors (time, frequency and GrO concentration) were proposed with two different values. This DE resulted in eight experiments.

Table 6.

DE for ball mill exfoliation.

Sample	Ball mill exfoliation
	Variable
	Frequency (Hz)	Time (min)	GrO (g)
GO-M1	7.0	5.0	0.025
GO-M2	7.0	8.0	0.1
GO-M3	10.0	8.0	0.1
GO-M4	10.0	8.0	0.025
GO-M5	7.0	8.0	0.025
GO-M6	10.0	5.0	0.1
GO-M7	10.0	5.0	0.025
GO-M8	7.0	5.0	0.1

Table 7 shows the results of the averages, ranges, SD and T_Cal of SSA, C (%), °2θ and “n” of the GO obtained by exfoliation with ball mill. The complete results can be found in Table Alll in Online Appendix A. It is observed that the highest average value of SSA obtained is 43.2 m² g⁻¹ and corresponds to samples obtained with a concentration of 0.025 mg of GrO. While the lowest value corresponds to an average SSA value of 28.5 m² g⁻¹ corresponding with 0.1 g of GrO. Concentration is the factor that more influences SSA result giving rise to the highest T_Cal (1.46), although its value does not exceed the critical value (2.45), so these differences are not statistically significant.

Table 7.

The textural, chemical and structural results obtained in ball milling optimisation.

	Ball Mill
	Variable	Frequency (Hz)		Time (min)		GrO (g)
SSA (m2 g−1)	Value	7	10	5	8	0.025	0.1
	Range	26.8–70.9	23.1–41.2	26.9–41.2	23.1–70.9	26.9–70.9	23.1–35.3
	Average	40.0	31.7	33.1	38.6	43.2	28.5
	SD	21.0	7.7	6.5	22.0	19.41	5.1
	TCal	0.74		0.48		1.46
	TCrit	2.45		2.45		2.45
C (%)	Range	38.7–39.2	38.7–39.3	38.7–39.2	38.7–39.3	38.7–39.0	38.7–39.2
	Average	38.9	38.9	38.9	38.9	38.8	39.0
	SD	0.2	0.2	0.2	0.2	0.1	0.2
	TCal	0		0		1.49
	TCrit	2.45		2.45		2.45
2θ (°)	Range	10.2–10.6	10.4–10.6	10.2–10.6	10.4–10.5	10.2–10.5	10.4–10.6
	Average	10.4	10.5	10.4	10.4	10.4	10.5
	SD	0.2	0.04	0.2	0.06	0.1	0.09
	TCal	1.03		0.09		1.27
	TCrit	3.18		2.45		2.45
n	Range	12.7–18.5	16.0–17.8.0.8	16.3–17.8	12.7–18.5	12.7–16.7	16.3–18.5
	Average	16.0	17.0	16.8	16.2	15.5	17.6
	SD	2.4	0.9	0.7	2.6	1.8	0.9
	TCal	0.81		0.46		2.03
	TCrit	2.45		2.45		2.78

In the case of the C (%) results, the average values are in the range of 38.8–39.0%. These values are very close. Also, GO obtained at different frequencies and milling times, present a T_Cal too small and are therefore shown as 0, revealing that there are no statistical differences in the C (%) in these samples when varying these experimental parameters. While the highest T_Cal (1.49) is obtained in samples synthesised with different amounts of GrO, being this parameter the most influential although it is not statistically significant, as in the case of SSA.

The results corresponding to the position, 2θ (°) show average values in the range of 10.4–10.5°, while for the number of stacked layers, "n", the average values are in the range of 15.5–17.6. In these results it can be seen that the highest T_Cal are those of the samples obtained with different amounts of GrO, both for the 2θ (1.27) and for (2.03). In both cases T_Cal are not statistically significant, although it is possible to highlight that the GrO amount has an influence on the number of layers of the obtained material. The higher the amount of GrO, the higher the number of layers.

It can be concluded that the most influential factor in the case of the ball mill exfoliation process is the amount of initial GrO used.

Figure 7a and b represent the Pareto chart and main effects plot, respectively, corresponding to the SSA obtained using the ball mill as exfoliation method. It is possible to see, in Pareto chart (Fig. 7a) that the amount of GrO, the milling frequency and time have a negative influence on SSA values obtained. The most negatively influential factors are the amount of GrO, as it is shown previously, and the interactions between time and frequency or amount of GrO. However, none of the effects studied is statistically significant on the results obtained for the SSA. According to the main effects plot (Fig. 7b), it can be seen that all variables studied, the amount of GrO, frequency and time decrease SSA values. High milling frequencies can lead to excessive fragmentation of GrO flakes, decreasing their SSA. While larger amounts of GrO generate higher friction and impact during milling and this may increase sheet fragmentation, resulting in lower SSA. The same effect is observed with an increase in milling time, as SSA decreases from approximately 39 to 33 m²:g⁻¹. Longer milling times allow more energy to be transferred to the material, promoting sheet breakage and leading to a reduction in specific surface area^101,102.

Fig. 7.

a and b Pareto chart and main effects plot respectively for SSA.

The Pareto chart and main effects plot for C (%) are depicted in Fig. 8a and b respectively. According to the Pareto chart (Fig. 8a), the factors that have the most favourable influence, increasing the C (%), are the amount of GrO and its interaction with frequency and time. While frequency and time separately do not influence the C (%) results. This is also observed in the main effects plot (Fig. 8b), where for both time and frequency horizontal lines of variation for C (%) are obtained when varying these two experimental factors. The carbon content remains almost constant with the milling time, around 38.9. Milling time within the range studied does not introduce significant chemical changes affecting the carbon content of graphene oxide. At the same way, the milling frequency does not significantly affect the carbon content, indicating that the frequency parameters within the studied range do not alter the chemical structure. On the other hand, it is observed that as the amount of GrO increases, the C (%) also increases. This may be due to greater friction and contact between particles, which provides more energy to break bonds of the oxygenated groups, increasing the C (%).

Fig. 8.

a and b Pareto chart and main effects plot respectively for C (%).

Glucose-assisted ball milling

The DE applied in this developed method is shown in Table 8. In the ball mill apparatus employed, time and frequency parameters can be selected. In this case, a new factor has been added, amount of glucose, with respect to the previous DE, therefore four factors (time, frequency and amount of GrO and glucose) are proposed with two different values.

Table 8.

DE for glucose-assisted ball mill exfoliation.

Sample	Ball mill-glucose exfoliation
	Variable
	Glucose (g)	Frequency (Hz)	Time (min)	GrO (g)
GO-MG1	3.5	7	5	0.025
GO-MG2	5.5	10	8	0.025
GO-MG3	5.5	7	5	0.1
GO-MG4	3.5	7	5	0.1
GO-MG5	3.5	7	8	0.025
GO-MG6	5.5	7	8	0.1
GO-MG7	3.5	10	5	0.025
GO-MG8	5.5	10	5	0.1
GO-MG9	3.5	10	5	0.1
GO-MG10	5.5	7	8	0.025
GO-MG11	5.5	10	5	0.025
GO-MG12	5.5	7	5	0.025
GO-MG13	3.5	7	8	0.1
GO-MG14	3.5	10	8	0.025
GO-MG15	5.5	10	8	0.1
GO-MG16	3.5	10	8	0.1

This DE resulted in sixteen experiments. The textural, chemical and structural results obtained in glucose-assisted ball milling optimisation are summarised in Table 9.

Table 9.

The textural, chemical and structural results obtained in glucose assisted ball mill optimisation.

Ball Mill with glucose	Variable	Glucose (g)		Frequency (Hz)		Time (min)		GrO (g)
	Value	3.5	5.5	7	10	5	8	0.025	0.1
SSA (m2 g−1)	Range	9.3–54.1	5.8–18.6	7.1–27.1	5.8–54.1	5.8–27.1	7.1–54.1	7.1–27.1	5.8–54.1
	Average	18.1	11.0	12.7	16.4	12.6	16.4	14.2	14.9
	SD	15.71	4.8	6.8	15.6	6.9	15.6	6.7	15.9
	TCal	1.21		0.61		0.63		0.11
	TCrit	2.31		2.23		2.23		2.26
C (%)	Range	31.0-40.9	33.5–39.2	31.0-39.8	33.5–40.9	31.0-40.9	33.5–39.8	31.0–37.2	36.7–40.9
	Average	37.7	37.2	37.1	37.8	37.3	37.6	35.5	39.4
	SD	3.1	2.3	2.9	2.5	3.1	2.3	2.2	1.2
	TCal	0.22		0.57		0.22		4.33
	TCrit	2.14		2.14		2.14		2.14
Position 2θ (°)	Range	10.4–11.2	10.6–11.2	10.4–11.2	10.5–11.2	10.4–11.2	10.6–11.2	10.4–11.0	10.9–11.2
	Average	10.9	11.0	11.0	10.9	10.9	11.0	10.8	11.1
	SD	0.3	0.2	0.3	0.2	0.3	0.2	0.3	0.1
	TCal	0.37		0.50		0.51		2.67
	TCrit	2.14		2.14		2.14		2.26
n	Range	7.6–14.6	7.9–15.0	12.7–18.5	16.0–17.8	16.3–17.8	12.7–18.5	12.7–16.7		16.3–18.5
	Average	12.6	12.4	16.0	17.0	16.8	16.2	15.5		17.6
	SD	2.1	2.0	2.4	0.9	0.7	2.6	1.8		0.9
	TCal	0.15		1.71		0.21		1.23
	TCrit	2.14		2.14		2.14		2.31

Table 9 shows the results of the average, ranges, SD and T_Cal of SSA, C (%), peak position 2θ (°) and “n” of the GO obtained by glucose-assisted ball mill exfoliation. The complete results can be found in Table AlV in Online Appendix A. The SSA results show average values in the range of 11.0–18.1.0.1 m² g⁻¹. Both the highest and the lowest values in this range correspond to samples obtained with higher and lower glucose content, respectively. Thus, the samples synthesised with different amounts of glucose have the highest T_Cal (1.21), although it does not exceed the critical value (2.31).

In the case of the results obtained for C (%), it is observed that the highest T_Cal (4.33) is obtained in samples with different initial amounts of GrO. This value exceeds the critical value (T_Crit=2.14), being a statistically significant effect on the C (%) of the samples exfoliated under these conditions. Samples with higher amount of GrO provide higher carbon concentration.

The average positions of the diffraction maxima are between 10.8 and 11.1 °. As in the case of C (%), the effect of the initial amount of GrO is statistically significant on the results, the T_Cal (2.67), exceeding the critical value (T_Crit=2.26). Samples with higher GrO give peaks further to the right, although in all cases the position of the resulting peak can be clearly assigned to a GO peak¹⁵.

A clear relationship can be seen between C (%) and peak position 2θ (°). The higher the C (%), the more the main peak (2θ) is shifted to the right. Higher C (%) indicates that this material has a lower proportion of oxygen and then it is a bit more reduced than others, so it is possible to indicate that in this case, the position of the peak shifted to the right could have a lower degree of oxidation. This justifies that in both cases the T_Cal is statistically significant.

The average values of “n” calculated are between 12.4 and 17.6, with the greatest differences being found in the case of samples obtained at different initial amounts of GrO (T_Cal=1.23) and at different frequencies (T_Cal=1.71). However, no experimental factor varied the results of “n” in a statistically significant way, as it did not exceed the critical value.

The Pareto chart and main effects plot of SSA for glucose-assisted ball mill exfoliation are shown in Fig. 9a and b respectively. According to the Pareto chart (Fig. 9a), the factors that most favour an increase in SSA are the interactions between time and frequency or amount of GrO, while an increase in glucose has a negative effect. However, none of the experimental factors had a statistically significant effect on the SSA of GOs obtained by this method. The main effects plot (Fig. 9b) show the variation of the SSA as a function of the different experimental factors studied. It is possible to observe that the frequency, time of milling and the amount of GrO generate an increase in the SSA of the samples obtained, while glucose has a negative effect.

Fig. 9.

a and b Pareto chart and main effects plot respectively for SSA.

The SSA increases with frequency, from approximately 13 to 16 m² g⁻¹. Higher frequencies facilitate better energy transfer to the particles, more effectively breaking up the layers and increasing the surface area. This trend is also favoured by longer milling times. As the time increases, there are more opportunities for the balls to exfoliate and disperse the layers, which increases the SSA by producing more graphene oxide thin films. A positive effect on SSA is also observed by increasing the initial amount of GrO. The SSA shows a slightly positive trend from approximately 14 to 15 m² g⁻¹. A higher amount of material could allow a better interaction and exfoliation, increasing the SSA. However, the SSA decreases significantly with increasing amount of glucose, from approximately 18 to 11 m² g⁻¹. Glucose may act as a reducing agent¹¹¹, interfering with the exfoliation of graphite oxide and promoting the formation of aggregates or interfering in the process of breaking down graphite oxide layers. The presence of glucose may introduce impurities or form a layer on the graphite oxide particles, reducing the exfoliation efficiency and decreasing the SSA of the final product.

Figure 10a and b, respectively, shows the Pareto chart and main effects plot of the results obtained for the C (%).

Fig. 10.

a and b Pareto chart and main effects plot respectively for C (%).

It can be seen, according to the Pareto chart (Fig. 10a), that the effects that most influence the C (%) of the samples obtained are GrO and the interactions between the frequency and time, besides, the amount of GrO has a statistically significant influence on the results. It is observed that an increase in time, frequency and the initial amount of GrO have a favourable effect, increasing C (%). The main effects plot (Fig. 10b) shows that milling frequency, time and amount of GrO increase C (%), while the amount of glucose has the opposite effect. Higher frequencies can increase exfoliation and reduction of the graphene particles, resulting in an increase of the carbon content. Prolonged time in the ball mill can promote the elimination of oxygenated groups by bonds breakage, thus increasing the relative carbon content. The carbon content increases with the amount of GrO, from approximately 35 to 39. An increase in GrO increases the carbon content, indicating a more efficient exfoliation and reduction process, due to the higher concentration of particles colliding with each other, causing greater removal of oxygenated groups through bond breakage. Finally, the presence of glucose may protect the graphene particles from oxidation during milling, thus preserving the carbon content.

Comparing methodologies

The four parameters studied in the different methodologies applied for the exfoliation of GrO to GO have allowed a deeper understanding of the mechanisms that take place during the exfoliation processes assisted by ultrasound or ball mill.

The aim of this study was to obtain GO with the highest SSA and a suitable carbon content in order to synthesise reduced materials (rGO or hydrogels/aerogels) in a simple way. Beyond this, this work gives us the possibility to evaluate all the studied parameters (SSA, C (%), peak position 2θ (°) and ‘n’) and to establish the best conditions to obtain the best results for each of them. In this way, the exfoliation can be directed to obtain the best conditions for the desired parameter. Figure 11 shows a comparison between the more favourable results obtained in the experiments that gave the best value for each parameter in each of the exfoliation method studied. In this figure, it is possible to see the effects on SSA, C (%), 2θ and “n”, due to the different variables associated with each process obtained by Designs of Experiments. SSA study (Fig. 11a) shows that this parameter is highly dependent on the method of exfoliation as there is a large variability in the results ranging from 10 to 70 m² g⁻¹. It is clearly evident that the methods using the ball mill, under the best conditions, result in materials with a higher SSA. In the case of graphite exfoliation, it is necessary to use exfoliating agents, such as glucose, when using ball mills as the exfoliation method¹¹², however, the results of this study show that the exfoliation of graphite oxide is more effective without exfoliating agent when using ball mills as the exfoliation method.

Fig. 11.

Best results for each parameter for each of the peeling methods studied: a SSA, b C (%), c position (2θ (°)), d number of layers.

For SSA survey, in the case of the ball mill, the variable that influences the final SSA of the GO materials obtained, is time. Time increase the SSA of the GO, due to the fact that the sample receives more energy to separate the layers and the effect of introducing oxygenated groups between the sheets. For exfoliation with ultrasound, samples with higher GrO concentrations have better SSA, perhaps because more amount of material does not break down the structures as much.

In the case of C (%), the aim was to obtain low oxidised materials (GO) that could be easily reduced, in order to achieve an adequate degree of reduction. So, GO materials with not very high content of oxygen-containing groups are needed for easier reduction, that is, the removal of these oxygen-containing groups¹¹³. Therefore, the highest C (%) was selected as the best results for C (%) for having the lowest percentage of oxygenated groups (Fig. 11b). The four methodologies studied do not greatly influence the results obtained since the variation found between the best values of the different methodologies are between 43.2% and 39.2%. It should be kept in mind that in ultrasound treatments the temperature of the solution has been kept constant at about 2 °C to avoid any effect (GO reduction) on the as-prepared materials, this has preserved the oxidized state of the materials. The results show that, in the case of ball mill treatments, the effect of the temperature rise that can occur inside the agitation container is negligible.

The time favoured the increase in C (%) in the ultrasound studies probably due to spot heating processes even though the processes have been done at low temperatures with an ice bath. In the ball mill studies, the best results have been obtained using glucose as the exfoliating agent and in this case the most influential parameters have been time and frequency and even more so the combination of both, the addition of glucose increases the carbon content, possibly acting as a reducing and protective agent against oxidation.

The position of the main peak near 10 ° reveals an oxidised structure^114,115, therefore, in this case, those with the lowest values close to 10 ° have been selected as the best results (Fig. 11c). Peak position is not greatly influenced by the different treatments performed for exfoliation. As in the case of C (%), this parameter was fixed in the previous step, GrO synthesis process, and is not so much affected by the subsequent exfoliation process. Thus, the range obtained for the best results of this parameter are from 9.8 to 10.4.

The peak position is also influenced by time in ultrasonic processes, being obtained better results with shorter times. As mentioned above in the C (%) study, time favours the appearance of hot spots in these processes which can remove a small amount of the oxygen-containing functional groups that have been introduced into the structure. For the methods studied using the ball mill, the lowest values for the peak position were obtained with the mildest conditions in all cases, which suggests that more aggressive conditions, in terms of time, frequency, GrO content and, also glucose content, can affect the proportion of oxygen-containing functional groups realising part of these groups, although this effect is negligible.

The parameter “n” related to the lowest number of GO layers resulting from each exfoliation method is shown in Fig. 11d. This parameter is the most influential parameter after SSA. The best value for the four methodologies are between 6.8 and 12.7 layers. Ultrasonic treatments produce materials with a lower number of layers but worse SSA, while in ball mill processes the number of layers is higher but the SSA is also higher.

In the case of bath and probe methods, short times favour to obtain materials with few layers, as occurs with SSA, because the structure is not broken as much by the effect of ultrasound. On the other hand, using the ball mill, the concentration of GrO is the common variable in both cases that favours to obtain materials with lower “n”. A lower concentration of GrO produces more exfoliated materials with a higher SSA and a lower number of layers.

The aim of this study was to obtain materials with the highest SSA and the highest carbon percentage, although, when carrying out the DE it has been observed that C (%) is a parameter that is not greatly influenced by exfoliation processes.

Ball mill exfoliation results in the highest SSA because during the process the material is further oxidised by the introduction of oxygenated functional groups favouring exfoliation. GrO exfoliation with ultrasonic probe is the more aggressive method, which causes more defects and fragmentation in the GO lamellae, resulting in smaller n and lower SSA. Ultrasonic bath exfoliation is a method that gives better results than probe exfoliation, it is a gentler method that generates fewer defects and fragmentation in the GO, so a future study combining ultrasonic bath and ball mill exfoliation could be considered to analyse its correlation with SSA⁷⁷.

These studies have demonstrated that the exfoliation of graphite oxide, obtained by the Tour method, is more effective, in terms of SSA, using a ball mill than ultrasonic exfoliation in water. In addition, the use of the ball mill directly, without exfoliating agent (glucose), was more effective, contrary to what usually happens with graphite exfoliation¹¹².

Characterization of the GO samples obtained under the optimal conditions for each methodology

As demonstrated in the previous section, SSA was identified as the most influential parameter among the methods evaluated. Consequently, the four samples obtained using the methods that produced the highest specific surface areas were selected for a more detailed analysis of their structural and morphological characteristics, employing FTIR, SEM and XRD techniques.

FTIR characterization

In Fig. 12, the FT-IR spectra of all the samples obtained using the different exfoliation methods under the best experimental conditions, for SSA parameter, are shown, together with the initial GrO (GO_initial). In all cases, baseline correction was performed using the ALS (Asymmetric Least Squares Smoothing) method, and normalization was carried out by dividing by the maximum intensity, in order to eliminate instrumental deviations and effects associated with differences in sample amount.

Fig. 12.

FT-IR spectra of GO samples obtained by different exfoliation methods under optimal conditions (GO_B3, GO_P1, GO_MG16 and GO_M5), together with the GO-initial, after ALS baseline correction and normalization to the maximum intensity.

The ν(C = O) band, located around 1720 cm⁻¹ and associated with carboxyl and carbonyl groups^116,117, appears more intense in the samples treated by milling without glucose and by ultrasonic bath (GO_M5 and GO_B3), indicating a greater preservation of oxygenated functional groups with these methodologies. In contrast, this band shows a very low intensity in the GO obtained by glucose-assisted ball milling (GO_MG16), clearly evidencing the reducing effect of glucose on these functional groups. In the case of the sample treated with ultrasonic probe (GO_P1), the observed intensity is lower than that obtained by ultrasonic bath, which can be attributed to the higher local temperature increase caused by direct contact between the probe and the solution, as well as to the more aggressive nature of the probe, whereas the ultrasonic bath constitutes a milder exfoliation method.

The signal around 1580–1620 cm⁻¹, attributed to ν(C = C) vibrations of sp² domains^118,119, is more intense in the samples exfoliated by milling methods, possibly because there is less fragmentation of the sheets, which preserves larger intact graphitic domains. In the samples obtained by milling, GO_M5 and GO_MG16, bands corresponding to water molecules adsorbed¹¹⁹ on the surface of their sheets are observed, δ(H–O–H)/δ(–OH) at 1560 and 1339 cm⁻¹. These can be associated with their higher degree of exfoliation and hydrophilicity compared to the US methodologies, which gives them a greater capacity to adsorb water from the environment.

These samples also exhibit bands associated with epoxy groups^119–121, ν(C–O–C)/ν(C–O) at 1071 and 972 cm⁻¹, whereas in the case of the two samples obtained by US a ν(C–O) band^121,122, associated with epoxy groups, is observed around 1025 cm⁻¹. This indicates that epoxy groups are preserved in the samples exfoliated by milling, possibly due to the lower aggressiveness of this method, which maintains larger GO sheets and preserves the cyclic epoxy groups intact. In the case of US-assisted methods, cavitation events are more aggressive and localized, which may lead, on the one hand, to a slight reduction of the materials (more pronounced in the US probe due to the local temperature increase), and on the other hand, to a possible opening of the epoxy rings, which are transformed into alkoxy groups, resulting in the appearance of the corresponding functional group band in the GO_B3 and GO_P1 samples.

It should be noted that the characteristic epoxy group band corresponding to the “breathing” of the ring, generally located at 1240–1250 cm⁻¹, does not appear in any of the cases, possibly due to the rigidity of the GO structures obtained after exfoliation. This is corroborated by the appearance of the epoxy ν(C-O) band at 972 cm⁻¹, at higher wavenumbers than those usually observed, indicating an increase in the rigidity of the corresponding bonds¹²¹.

Overall, a general tendency of the samples to undergo reduction after the exfoliation process is observed. This tendency is more pronounced in the case of the US probe, followed by the ultrasonic bath, and finally by the ball-milling methods, during which the energy is distributed more homogeneously over the material. Interestingly, in this case, the milling methods generate samples in which bands corresponding to epoxy groups appear, which were not present in the initial GO, possibly as a result of the incorporation of ambient O₂.

SEM characterization

Figure 13 shows the SEM micrographs of the GOs obtained by the different exfoliation methods under the optimized experimental conditions, for SSA parameter. A clear correlation can be observed with the textural and structural results discussed in the previous sections.

Fig. 13.

SEM images for GO samples obtained under the optimal conditions for each methodology: a without exfoliation; b US bath; c US probe; d ball mill; d glucose assisted ball mill. a GO_initial, b GO_B3, c GO_P1, d GO_M5, e GO_MG16.

In the sample corresponding to GO_B3 (Fig. 13b), relatively uniform sheets can be observed, with a low degree of fragmentation and slightly pronounced surface wrinkles. The layered structure is partially separated, confirming an efficient and controlled exfoliation process. This is consistent with the intermediate-to-high SSA values obtained and corroborates that exfoliation by US bath promotes exfoliation with a lower introduction of structural defects.

In the case of the ultrasonic probe treatment (Fig. 13c), sheets with greater roughness, the presence of deformations, wrinkles, fractured areas, and particles with a tendency to re-stack are observed. This effect indicates that intense and localized cavitation is generated during exfoliation as a consequence of the aggressiveness of this method. This morphology confirms the lower SSA and C (%) values obtained, due to excessive defect generation and partial loss of oxygen-containing groups, with possible re-stacking of the sheets.

Figure 13d shows the SEM image corresponding to sample GO_M5, subjected to ball milling exfoliation treatment. Thinner and more fragmented sheets are observed, a more heterogeneous dispersion, greater interlayer spacing, and surfaces with irregularities, which is consistent with the higher SSA values obtained for this method. However, structural defects associated with repetitive mechanical impact are also identified, which is related to the relatively high number of layers in this sample. Compared to the ultrasound-based methods, ball milling provides more effective exfoliation in terms of SSA, although at the cost of introducing a higher density of defects.

In the optimal sample associated with the glucose-assisted ball milling exfoliation method (GO_MG16, Fig. 13e), a more compact structure and the presence of agglomerates can be observed, which suggests that glucose partially interferes with the separation of the sheets. This observation is consistent with the intermediate SSA values and the increase in C (%) detected for this method, where glucose appears to promote slight chemical reduction while at the same time limiting complete exfoliation.

XRD characterization

Another important parameter obtained from XRD analysis is the crystallite size, commonly estimated using the Scherrer equation, which reflects the extent of fragmentation resulting from the various processing methods and instrumentation used.

Figure 14 presents the crystallite sizes of the four samples obtained under the most favourable conditions with respect to SSA. It is observed that milling-based exfoliation leads to larger particle sizes (10.8 and 10.2 nm for GO_M5 and GO_M16, respectively). In the case of experiments performed using ultrasonic treatments, smaller crystallite sizes were obtained (6.6 and 8.6 nm for GO_P1 and GO_B3, respectively), due to the more energetic and localized nature of these processes, as previously discussed. These results are consistent with those previously obtained for the number of layers, as this parameter shows a proportional behaviour (d = 0.8 nm for almost all GO samples). SEM images further support this trend, revealing more agglomerated particles in samples subjected to ball-milling treatments, whereas samples treated with ultrasonic probes exhibit a more homogeneous and dispersed morphology.

Fig. 14.

XRD crystallite size for GO samples obtained under the optimal conditions for each methodology .

Therefore, the methods showing the highest SSA values were selected as the most representative within each methodology. Table 10 summarizes the applied experimental conditions for each case, together with a comparison with data reported in previous bibliographic studies.

Table 10.

Optimal results and conditions of each methodology used, and comparison with previous works reported in literature.

Experiment (synthesis method)	This work
	Experimental conditions					Results
	Medium	Power (%)/frequency(Hz)	Time (min)	GrO (g)	SSA (m2 g−1)	C (%)	Position 2θ (°)	n
GrO-initial (Tour)	Water	–	–	–	13.96	39.4	10.74	18
GO-B3 (Tour)	Water	–	30	2	29.79	39.9	10.44	10.18
GO-P1 (Tour)	Water	30	30	2	10.07	41.2	9.85	7.35
GO-M5 (Tour)	Air	7	8	0.025	70.93	39.0	10.46	12.74
GO-MG16 (Tour)	Glucose (3.5 g)	10	8	0.1	54.06	39.3	11.00	12.72

Synthesis method GrO/Exfoliation to GO	Previous works
	Experimental conditions				Results
	Medio	Power/frequency	Time	GrO	SSA (m2 g−1)	C, O (%); C/O	Position 2θ (°)	n	References
Hummers/Ball Mill	Acetone	300 rpm	5–50 h	1.4 g	55.5–2.23	O: 21.7–33.5	11.4–23.8	2–3	101
Modified Hummers/US	Water	20.4 kHz	1–8 h	3 mg mL−1	–	C: 46.03–47.96; O: 51,15–49.11.11	12	–	123
Hummers/US Probe	Water	42 kHz; 240 W	5–30 min	–	–	C/O: 1.337–1.753	9.062–9.162	< 2	77
Hummers/US Bath	Water	42 kHz; 160 W	1–5 h	–	–	C/O: 1.520–1.094	8.841–8.863	< 2
Modified Hummers/US Probe	Water	20 kHz; 750 W	5–20 min	20 mg mL−1	–	–	~ 13–14	> 30; < 5	124
Modified Hummers/Ball Mill	Water	500 rpm	10–40 h		–	C: 46.31–46.8	10.7	> 10 (10 h) fully exfoliated (30 h)	96
Hummers/Ball Mill	Water; NMP	300 rpm	6 h	300 mg	–	C: 88.23; O: 10.42 (NMP)	10.9, 25.8 (in NMP)	–	102
Hummers/US Bath	Water; NMP	–	8 h	300 mg	–	C: 69.18; O: 27.01 (in NMP)	10.88 (in NMP)	–
Hummers/US	Water	200 W	30 min	2 g L−1	47.7	–	9.8	–	125

By comparing the results obtained in this work with those reported in previous studies on the mechanical exfoliation of GrO, a clear lack of research related to GrOs synthesised by the Tour method becomes evident. Most of the works described in the literature address the exfoliation of GrO synthesised by the conventional Hummers’ method. Moreover, many of the published studies focusing on exfoliation processes do not include key structural parameters, such as the number of stacked layers or the specific surface area, which are essential to correlate the exfoliation conditions with the final properties of GO.

This study fills that gap by providing a systematic analysis of the mechanical exfoliation of Tour-derived GrO, using both ultrasound-assisted and ball-milling methods, offering direct experimental evidence of how processing parameters influence SSA, "n", and C (%). The results obtained in this work not only contribute to a deeper understanding of the behaviour of highly oxidised GrO during exfoliation processes, but also establish a basis for optimising GO synthesis through scalable and environmentally friendly mechanical methods.

Graphene materials with high SSA have a high applicability due to this characteristic. SSA refers to the effective surface area under unit mass that can provide potential surface adsorption sites, these materials can therefore be employed in applications where the adsorption process is compromised. It can be used in applications where gases have to be retained such as hydrogen adsorption^49,126,127. That is now a very active area of research since hydrogen is an energy carrier. These methods can also be used to retain other harmful gases^128,129.

As a retention system, GOs with high SSA can also be used for the adsorption of elements or compounds in solution^130,131. A major application of these developments is the removal of possible reactive elements in solution from accidents or normal processing in nuclear power plants^132,134,134.

It can also be used in electrochemical processes such as those found in supercapacitors where the higher the SSA, the more energy storage capacity it will provide^9,60,135,136. Moreover, SSA is critical in other applications, such as photovoltaic¹²⁷ cells or catalytic processes¹³⁷, where performance is directly determined by this factor.

Conclusions

In this work, a DE approach was applied to carry out a systematic and rigorous study of the relationship between the experimental exfoliation conditions and the resulting structural and textural characteristics of GO synthesised by the Tour method. The exfoliation methodologies analysed were selected for their simplicity, low energy requirements, and environmentally friendly character. Four mechanical exfoliation methods were evaluated: two ultrasound-assisted and two ball-milling-based. In all of them, the exfoliation times did not exceed 90 min, reaching as little as 5 min in the case of the milling methods, which represents a great advantage compared to similar methodologies in which exfoliation times on the order of several hours (up to 30–50 h) have been reported.

Among the methods studied, ball milling without additives produced the highest SSA (70.9 m² g⁻¹) and the lowest number of stacked layers, while glucose-assisted ball milling yielded intermediate SSA values (54.1 m² g⁻¹). In contrast, ultrasound-assisted exfoliation generated lower SSA values (6–30 m² g⁻¹) due to greater sheet fragmentation. In the case of the US probe, an increase in C (%) was observed and corroborated by the FTIR results, caused by local temperature rises, leading to partial reduction, while glucose-assisted milling achieved C (%) values of up to 31.0, revealing its protective effect against reduction. XRD analysis confirmed that variations in exfoliation conditions directly affected the diffraction peak position, number of stacked layers (n = 7–17) and crystal size (6.6-10.8 nm), confirming a clear correlation between the experimental parameters and the structural integrity of the material. FTIR analysis further indicated that the ultrasonic probe promotes a higher degree of reduction of oxygenated functional groups due to spot heating. These results demonstrate that both the structural and chemical properties of GrO can be tuned by adopting different exfoliation strategies and adjusting their experimental parameters.

This study provides rigorous and quantitative evidence of the relationship between exfoliation conditions and the structural evolution of the obtained GO, establishing a solid framework for its synthesis and controlled optimisation through green, scalable, and application-oriented mechanical methods. The results contribute to advances in the production of GO with tunable properties, facilitating its application in GBMs intended for energy storage, environmental remediation, and the development of advanced functional materials.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

“Conceptualization, Hanna Bukovska and M. Belén Gómez-Mancebo; methodology, Hanna Bukovska; software, Hanna Bukovska; formal analysis Hanna Bukovska, Fernando García Perez, Susana Fernández, Iván Sánchez, Manuel Fernández; investigation, Hanna Bukovska and M. Belén Gómez-Mancebo; resources, Javier Martínez, Isabel Rucandio and M. Belén Gómez Mancebo; writing—original draft preparation, Hanna Bukovska.; writing—review and editing, Hanna Bukovska, M. Belén Gómez-Mancebo and Isabel Rucandio; supervision, M. Belén Gómez Mancebo; project administration, M. Belén Gómez Mancebo, Javier Martínez and Isabel Rucandio. All authors have read and agreed to the published version of the manuscript.”

Funding

This research was supported by REGRAP-2D, Grants PID2020-114234RB-C21 funded by MCIN/AEI/10.13039/501100011033. During the preparation of this work the author(s) used “DeepL“in order to check the grammar. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

Data availability

Data is provided within the manuscript.

Declarations

Competing interests

The authors declare no competing interests.