1. Introduction

Aqueous graphene oxide (GO) dispersions exhibit a self-assembled liquid crystalline phase due to their highly anisotropic shapes with high aspect ratios [1, 2]. Shen et al. reported that aqueous GO dispersions with low concentrations have extremely large Kerr coefficients in the isotropic phase and as a result, the alignment of the GO particles in the dispersions can be switchable by applying weak electric fields [3–5]. GO dispersions with large Kerr coefficients may be highly useful for electro-optical applications, such as low-power displays. Note that low-power display technology is being intensively investigated by many researchers because it is an essential component for future wearable IT devices.

In birefringence-type displays, such as liquid crystal displays, the optical retardation, which is defined as the product of the light-path length (d) and the birefringence (Δn), should be higher than the half of the light wavelength (λ/2) to obtain the maximum optical efficiency [6]. Since Δn is directly proportional to the concentration of GO in aqueous GO dispersions, one may need to use a GO dispersion with high concentration in order to develop an electro-optical device. However, Shen et al. reported that nematic GO dispersions with high GO concentrations do not respond to the application of electric fields due to strong interparticle frictional interactions [3]. As the GO concentration increases, the spacing between neighboring GO flakes decreases and the rotational motion of GO particles can be easily hindered. Since the electrical switching of GO dispersions is achieved through the rotational motion of individual GO particles, the increase in GO concentration sensitively influences the electrical sensitivity of the GO dispersion. This can be a serious drawback for electro-optical devices using GO dispersions. Intuitively, we can assume that the GO particle size will influence both the friction of the rotational motion of the individual GO particles and the electro-optical sensitivity.

In this study, we investigated the influence of the GO particle size on the electro-optical sensitivity of aqueous GO dispersions. The electro-optical sensitivity of GO dispersions with high concentrations was significantly improved by reducing the mean GO size. In order to clarify the underlying mechanism, phase diagrams were obtained as functions of particle size and concentration and the viscosity was analyzed for various samples. Not only the optical birefringence, but also the dynamic response time, was significantly improved by optimizing the mean size of GO particles in the electro-optical switching of GO dispersions.

2. Experiments and results

An aqueous GO dispersion was prepared by using the Hummers method [7].

A Scanning Electron Microscope (SEM) was used to measure the GO particle size (D). The samples for the SEM measurement were prepared by spin coating the GO dispersion on silicon substrates and drying them at room temperature. The GO dispersion had a mean particle size of 7.95 µm, which is quite large. We can reduce the mean particle size by ultra-sonicating the GO dispersion, because GO particles are shattered during inter-particle collisions and collisions with the wall of the bottle [8]. As the sonication time increased using an ultra-sonication bath (SD-250H by Mujigae Co., Korea), the mean size of the GO particles exponentially decreased, as shown in Fig. 1(a), in which the x-axis is in logarithmic scale. We selected four GO dispersions with different particle mean sizes ranging from 7.95 µm to 0.075 µm, as shown in Fig. 1(b).

 

Fig. 1 (a) Mean GO size as a function of ultra-sonication time. (b) GO size distributions for four selected GO dispersions and corresponding SEM images.

Download Full Size | PDF

As the GO concentration in aqueous dispersions increases, the phase changes from isotropic to biphasic and from biphasic to nematic [3, 9]. We examined the phase diagram for the four selected samples by observing the optical birefringence textures [3], and the result is shown in Fig. 2(a). The phase transition concentration depends on the particle size. As the particle size decreases, the phase transitions occur at higher concentrations, as indicated in Fig. 2(a). The GO dispersion with the large flake size of 7.95 µm exhibited the nematic phase even at 0.2 wt%, whereas the GO dispersion with the mean size of 0.075 µm was biphasic even at 3.5 wt%. This is well explained by the Onsager type excluded volume theory, which states that the phase transition concentration sensitively depends on the aspect ratio of the particles [10–12]. The width of the biphasic region partially depends on the polydispersity of the particles [11]. Table 1 shows the specifications of the samples on the border of the isotropic to biphasic transition [red data points in Fig. 2(a)]. These samples were in the isotropic phase, which means no self-ordering and no spontaneous birefringence.

 

Fig. 2 (a) Phase diagram of GO dispersions as functions of the mean GO particle size and concentration. (b) Viscosity of the four samples on the isotropic to biphasic transition line. The inset curve is for a fixed GO size and the inset data were taken from reference [2] for comparison.

Download Full Size | PDF

Table 1. Mean size and concentration for the GO dispersion samples on the isotropic to biphasic transition (See Fig. 2(a))

View Table

We measured the viscosity of the samples shown in Table 1 by using a micro-Ostwald viscometer with an inner capillary diameter of 0.4 mm from SI-Analytics GmbH (Germany). Figure 2(b) shows the viscosity of the samples as a function of GO concentration. The viscosity only slightly increases with increasing concentration, as indicated in Fig. 2(b), which shows a clear contrast to the result shown in the inset. The inset in Fig. 2(b) is from reference [2], in which the viscosity of GO dispersions with fixed particle size (3.2 μm) was measured as a function of GO concentrations and it was found to increase exponentially with increasing GO concentration. The viscosity of the 1 wt% sample in the inset data is approximately three orders of magnitude larger than that of sample S_C, which has also a concentration of 1 wt%. The difference originates from the different particle size. Note that all the samples were in the isotropic phase, whereas the 1wt% sample presented in the inset was in the nematic phase. The nematic phase appears because of the interparticle steric interaction that hinders the free rotation of the GO particles and causes their packing. Usually, the spacing between neighboring particles decreases as the concentration increases, which results in an increase in the rotational friction of the particles and in the viscosity. However, for smaller particles, although the spacing between neighboring particles is small, the rotational friction is still low enough to allow free rotation, which gives rise to a high-concentration isotropic phase. Thus, the viscosity is much lower for small-sized GO dispersions.

Subsequently, we measured the electro-optical response of GO dispersions with different flake sizes and concentrations. The cell used in our experiment is shown in Fig. 3(a). Copper wires with a thickness of approximately 150 μm were interdigitatedly placed on a bare glass substrate, as shown in Fig. 3(a), and another bare glass substrate was placed on top of it. The cell gap was about 1 mm, which was sustained by a space holder and was filled with the GO dispersion. The cell was sandwiched by two crossed polarizers. The distance between two neighboring copper wires was approximately 1 mm. The odd copper wires and the even copper wires were connected to the ground and to an electric signal with 10 kHz, respectively. A light beam from a halogen lamp was filtered using a narrow band-pass filter (λ = 550 nm). In our previous work, we have found that the wavelength dispersion of GO dispersion is relatively weak compared to usual liquid crystals {see Supplementary Fig. 7(b) in reference [3]}. Hence, the choice of a specific wavelength does not influence the overall conclusion of the study. The light intensity was measured as a function of the amplitude of the signal by using a photo-diode. The effective birefringence (Δn) was calculated as

 

Fig. 3 (a) Photographs of the cell used in the experiment (the GO sample pictured in the cell is S_C); the left image is the ‘field off’ state and the right image is the ‘field on’ state under crossed polarizers. Arrows denotes the directions of polarizers. (b) Birefringence as a function of applied electric field for the GO dispersions S_A to S_D. (c) The birefringence at 20 V/mm for GO dispersions with different GO sizes.

Download Full Size | PDF

Figure 3(a) shows the off and on states of the cell filled with the S_C GO sample (1 wt% concentration and 0.51 µm flake size). As the GO sample was initially in the isotropic phase, the cell looked dark under the crossed polarizers, as shown in the left image in Fig. 3(a). A square-wave signal (10 kHz, 10 V) was applied to turn it on. In the ‘on’ state [the right image in Fig. 3(a)], the GO particles aligned and exhibited electric field-induced birefringence (EIB), so the cell appeared bright.

The EIB was measured for the S_A to S_D samples, which are on the border of the isotropic to biphasic transition for each size, and the result is shown in Fig. 3(b). Although the size was different, the EIB increased with increasing concentration. Figure 3(c) shows the EIB at 20 V/mm as a function of GO concentration for different flake sizes; the dotted lines are used to demonstrate the curve trend and do not represent experimental data. Overall, the curves exhibit a similar trend; at low concentrations, the birefringence at 20 V/mm increases with increasing concentration and then it decreases above a certain concentration. The arrows indicate the isotropic to biphasic transition concentration for each GO size, which correspond to S_A to S_D. The electric sensitivity increases up to a low-concentration biphasic state. When no field was applied in the thin cell, the biphasic dispersion appeared dark and the electro-optical response could be obtained. As the concentration increased further to high-concentration biphasic and nematic phases, the electrical sensitivity decreased rapidly. The peak concentration for the maximum birefringence increased as the mean GO size decreased, as indicated in Fig. 3(c). The maximum birefringence also increased as the mean GO size decreased to 0.51 µm, and the maximum birefringence was obtained for the 2 wt% GO dispersion with a mean size of 0.51 µm. However, further decrease in the particle size reduced the maximum birefringence, and the GO dispersion with mean size 0.075 µm had a lower value of maximum birefringence compared with the GO dispersion with mean size 0.51 µm. As the particle size decreased, the interparticle frictional interaction was reduced, but at the same time the electrical polarizability decreased as well. When the GO size is smaller than about 0.1 μm, the reduction in electrical polarizability predominantly influences the electro-optical sensitivity. As a result, the maximum birefringence was obtained in the GO dispersion with a mean size of approximately 0.5 μm.

The dynamic response time is another important parameter for electro-optical applications [6]. We measured the dynamic electro-optical response for all samples of Table 1. We applied a square-wave electric field with 10 kHz and 10 V/mm, and the rising and falling response curves were recorded when turning the signal on and off, respectively. The results are shown in Figs. 4(a) and 4(b). For the rising curves, the dynamic response becomes faster as the mean GO size decreases from 7.95 μm to 0.51 μm, but it is slower for the S_D sample than for S_C. On the other hand, the falling response displays a clear trend of an increasing falling speed as the mean GO size decreases.

 

Fig. 4 (a)-(b) Dynamic electro-optical responses during switching on and off, respectively. (c) Time constants for each dynamic response and the applied single exponential curve fits.

Download Full Size | PDF

In order to clarify the observed dynamic behavior, the response curves in Figs. 4(a) and 4(b) were fitted with a single exponential curve as

3. Conclusion

We investigated the effect of the mean GO size on the electro-optical sensitivity of aqueous GO dispersions and concluded that the sample with mean particle size around 0.5 μm showed the best performance regarding both the birefringence and the dynamic response. By reducing the mean GO size to 0.5 μm, we could obtain a 2 wt% GO dispersion that was electrically switchable. The birefringence for the sample with 2 wt% concentration and 0.5 μm mean size was about 6.5 times higher than that of the sample with 0.1 wt% concentration and 7.95 µm mean size. The fastest response time was obtained for the sample with a mean size of approximately 0.5 μm and further decrease in size increased the rising response time. The results can be useful for electro-optical applications using GO dispersions.