1. Introduction

Liquid crystal (LC) materials [1,2] have been used in various consumer electronic (LCDs [3] and webcams [4]) and specialized electronic (polarizer [5] and variable optical attenuator, [VOA] [6,7]) devices. The absence of mechanical movement, low cost, low voltage and power consumption are among the most attractive features of LC-based components. Very often, however, those components are perceived as being slow. In reality, the speed requirements depend upon the type of application, ranging from submilliseconds (shutters), milliseconds (LCDs), tens of milliseconds (network reconfiguration), and up to seconds (privacy windows).

It is well known that the speed of LC-based devices is typically defined by the natural (or free) relaxation time ${\tau }_{\text{off}}\approx \gamma d^2/(K{\pi }^2)$ of the deformation of their director (average orientation of long molecular axes of the LC [1,2,8], where $\gamma$ is the rotational viscosity, $K$ is the elastic constant and $d$ is the thickness of the nematic LC (NLC). In the same time, the typical excitation times ${\tau }_{on}$ of those materials are defined as ${\tau }_{\text{on}}={\tau }_{\text{off}}/({(V/V_{\text{th}})}^2-1)$), where $V$ is the excitation voltage and $V_{\text{th}}$ is the director’s reorientation threshold voltage [2,8]. Therefore, while the ${\tau }_{\text{on}}$ may be controlled by the choice of the voltage $V$, the ${\tau }_{\text{off}}$ is difficult to control dynamically for a given type of NLC material, cell geometry, and operation temperature. Thus, the duration of the full cycle of excitation and relaxation in standard NLCs is mainly defined by the free relaxation time ${\tau }_{\text{off}}$.

To overcome the abovementioned speed limitation, “dual-frequency” NLC (DF-NLC) mixtures were introduced, which have positive dielectric anisotropy $\mathrm{\Delta }\epsilon >0$ at low frequencies (e.g., at 1 kHz) and negative dielectric anisotropy $\mathrm{\Delta }\epsilon <0$ at higher frequencies (e.g., for $>50\mathrm{kHz}$ at room temperature). Such mixtures were successfully used, among others, to build fast VOAs [9] and phase modulators [10]. However, usually the uniform NLC cells (both ordinary and DF controlled) require polarizers to obtain intensity modulation. To obtain transmission modulation without polarizers, polymer dispersed LC (PDLC, [11]) and polymer stabilized LC (PSLC) compositions were introduced with ordinary NLCs [12] as well as with DF-NLCs [13]. In those PSLC materials, the polymer network (bundles, walls, and aggregates) is dispersed in the volume of the NLC. Later, surface polymer stabilized NLCs (S-PSLCs) were introduced [14–16] with the promise of better control and stability. Those material compositions were using ordinary NLCs. However, in contrast with PSLCs, the polymer matrix here was attached to the surface of the cell substrates and only partial interpenetration was allowed between the NLC and the polymer network. While the obtained response times in all above-mentioned cases were noticeably faster (almost by an order of magnitude compared to pure NLCs), the modulation speed and contrast were still rather limited. We were thus interested to see if we could further improve our capacity to configure such components (see hereafter) as well as their performance by using DF-NLCs in the same (S-PSLC) geometry. For the sake of shortness we shall further call them as S-PS-DF-NLC.

Thus, in the present work, we describe the fabrication (photoelectric configuration) and experimental characterization results of S-PS-DF-NLC cells. Given the large variety of possible fabrication processes, we limit this part to one or two examples, while the electro-optic characterization of obtained cells is presented in more detail. The scattering modulation contrast, transition times, and polarization dependences are compared for the cases of excitation followed by natural relaxation as well as for various excitation sequences. As it will be shown hereafter some of the key technical parameters of obtained cells may be in “conflict” and a corresponding choice must be done as trade-off when using DF-NLCs in S-PSLC configuration.

2. Materials and Methods

Two commercially available glass substrates (already covered, from one side, by transparent conductive indium tin oxide, ITO, layers) were used to build sandwich-like cells containing our S-PS-DF-NLC. Each substrate was first covered by a layer of $\approx 50\mathrm{nm}$ thick polymide (PI-150 from Nissan). Note that we usually apply this polymide to obtain a planar (parallel to the cell substrates) aligning layer through the well-known mechanical rubbing procedure [17]. However, in the present work, we did not rub the PI-150 layers to eliminate the preferential alignment direction and thus to obtain polarization-independent operation. Note that we had two reasons for the use of PI: to have similar substrate (as those used in previous studies) and to provide good uniformity of the next thin layer. Indeed, a layer ($\approx 600\mathrm{nm}$ thick) of reactive mesogen (RM) was cast onto each PI layer by spin-coating procedure. Then those RM films were dried and annealed on a hot stage at 60°C for $\approx 90\mathrm{s}$. We have used two such substrates to fabricate the sandwich-like cell by the “drop-fill” technique. This was done without polymerizing the RM layer by UV light (usually it is done right after the deposition of the RM layer). Thus, one of the substrates was positioned horizontally (with the RM layer facing up) and the droplet of the DF-NLC (MLC2048, from Merck) was dispensed onto the surface of the RM layer. The periphery of the substrate was covered by UV curable adhesive walls (AC A1432-Q-S2 from Addisson) containing glass spacers ($\varnothing \approx 50\mathrm{\mu m}$) to ensure the desired thickness of the cell. Then the second RM-covered substrate was pressed (with the RM layer facing down) on the drop of DF-NLC and the peripheral adhesive was cured (by a UV lamp with spectra between 300 and 450 nm, intensity of $\approx {10\mathrm{mW}/\mathrm{cm}}^2$, irradiation time $\le 10\min$) while protecting the central part of the cell (the zone, which was later used for electro-optic tests) by a mask. An electric field ($110{\mathrm{V}}_{\mathrm{RMS}}$, AC of SIN form at 1 kHz or 70 kHz, see hereafter) was applied to the cell during these 10 min to allow reorientation and partial interpenetration of DF-NLC and RM molecules, prior to the total curing of RM (over the entire surface of the cell) by the same UV lamp during 40 min. After this polymerization step the electric field was switched off (for the sake of shortness, we shall further call the above-described process as cell programming). However, since the polymerization was performed in the nonequilibrium state (field-induced alignment, interdiffusion, and photopolymerization), the cells were subjected to a final voltage pulse (e.g., an AC voltage of $U=110\mathrm{V}$ and 1 kHz frequency during 1 s) to obtain a stabilized (ground-state) material morphology before their electro-optic study. The five major steps of cell fabrication are summarized in Fig. 1.

 

Fig. 1. Schematic presentation of major steps of the fabrication of S-PS-DF-NLC cell (see text for details).

Download Full Size | PDF

3. Experimental Setup and Process

To monitor the programming process, we have used a standard polarimetric setup (Fig. 2). Namely, the S-PS-DF-NLC cell was placed between crossed polarizer and analyzer (Glan prisms) and the transmission of the probe beam (CW He–Ne laser, operating at 632.8 nm) through the analyzer was monitored dynamically by using a photodetector (the half-wave plate $\lambda /2$ and the diaphragm were added later, for electro-optic scatter measurements only). At the programming stage, the probe beam was at normal incidence on the cell.

 

Fig. 2. Schematics of the general experimental setup used for the “programming” and for the electro-optic study of the haze and angular dependence of light scattering of S-PS-DF-NLC cells. The half-wave plate $\lambda /2$ and the diaphragm were used later for electro-optic tests only (see hereafter).

Download Full Size | PDF

The original “drop-fill assembled” cells were transparent in their ground state (before the application of the electric field and before the curing of the RM layers), while having a planar-aligned director with large domains delimited by disclination (orientational defect) lines. Thus, the initial transmission of the probe beam was relatively high, which was defined by the local effective birefringence ${\mathrm{\Delta }n}_{\text{eff}}$ and the corresponding phase delay $\mathrm{\Delta }\varphi =d\mathrm{\Delta }{n}_{\text{eff}}{2\pi /\lambda }_0$, where ${\lambda }_0$ is the wavelength of light in vacuum and $\mathrm{\Delta }{n}_{\text{eff}}\equiv n_e-n_{\perp }$ is the effective optical birefringence of the NLC. It is well known (see, e.g., [10] and references therein) that the addition of a strong electric field (e.g., $U=110\mathrm{V}$; note that all voltages will further be expressed in RMS) reorients the director to become perpendicular to the cell substrates (often called homeotropic alignment) if the electrical signal has relatively low frequency $f$ (e.g., 1 kHz), at which, the $\mathrm{\Delta }\epsilon (f)>0$. In contrast, the alignment of the director is “forced” to become planar if the frequency is high (e.g., $U=110\mathrm{V}$ at 70 kHz), at which the $\mathrm{\Delta }\epsilon (f)<0$. In the case, if there is no rubbing (preferential direction), this planar alignment is strongly nonuniform. Indeed, in both cases, we observed significant reduction of the transmitted probe signal; in the first case (excited homeotropic state), because of $\mathrm{\Delta }{n}_{\text{eff}}=0$, and in the second case (planar-excited state), because of the strong scattering of light. During the above-mentioned electric-field-induced reorientation, the molecules of MR and DF-NLC underwent mutual interdiffusion [14–16]. Then, the broadband UV light source was turned on during 40 min to polymerize the RM. Then, the UV light was switched off first and then the electric field was switched off too. The final “remnant” transmission was much lower compared to the initial state.

The cells were then subjected to electro-optic measurements (after their electrical “stabilization” or “annealing” by applying $U=110\mathrm{V}$ at 1 kHz during 1 s). For the following measurements, we have used the same experimental setup that is schematically presented in Fig. 2. At this stage we have removed the analyzer and added the half-wave plate $\lambda /2$ to change the orientation of the linear polarization of the probe beam (vertical or horizontal). The diameter of the diaphragm (also added at this stage), which is positioned just in front of the detector, was $\varnothing \approx 0.5\mathrm{mm}$, and its distance from the cell was 70 cm (providing a total acceptance angle of $\theta \approx 0.04\mathrm{rad}$). We have measured the transmission of the probe beam in different states of the cell. Thus, a low-frequency (1 kHz) voltage was applied to the sample to obtain excited homeotropic state. Then the voltage was switched off to observe the natural relaxation of the cell to its ground state. Alternatively, a high-frequency voltage (e.g., at 80 kHz) was applied to the cell immediately after the application of the voltage at 1 kHz to obtain planar-excited state instead of natural relaxation. All those measurements were made for two linear polarizations of the probe beam, parallel and perpendicular to the horizontal plane.

4. Stationary Mode Characterization

Before describing the detailed results of our electro-optic measurements, we can “visually” appreciate the performance of obtained samples. Figure 3(a) shows the example of a cell, which was programmed with $U=110\mathrm{V}$ at 1 kHz. As one can see, there are two scattering states (left picture: ground state with $U=0\mathrm{V}$, and right picture: excited planar state, with $U=80\mathrm{V}$ at 80 kHz). The cell is positioned on the top of the bar code pattern and pictures are recorded thanks to the reflected natural light without polarizers. This was made possible since the polarization dependence of our cells was noticeably lower (see hereafter) compared to the case when rubbed PI was used [14]. As we can see [Fig. 3(a), in the center], the application of the low-frequency control (or excitation) voltage $U=110\mathrm{V}$ at 1 kHz reduces the light scattering. The analyses of the intensity distribution shows [Fig. 3(b)] that the modulation depth is noticeably stronger between the excited homeotropic [Fig. 3(a), center] and excited planar [Fig. 3(a), right] states.

 

Fig. 3. Qualitative demonstration of (a) cell’s scattering in its ground state (left picture, $U=0\mathrm{V}$), when subjected to $U=110\mathrm{V}$ at 1 kHz (central picture) and when subjected to $U=80\mathrm{V}$ at 80 kHz (right picture). (b) The comparative histograms for three key cases (vertical arrows show the scanned zones). The cell was programmed with 110 V at 1 kHz.

Download Full Size | PDF

Similar cells were used to measure the angular distribution of the scattered light. For this experiment, the setup of Fig. 2 was used (without the analyzer) with the photodetector and diaphragm turning around the position of the cell (all elements staying in the horizontal plane). As one can see (Fig. 4), the probe’s divergence (originally $\approx {10}^{-3}\mathrm{rad}$) is increased by 3 orders of magnitude and it is almost twice larger for the cell that was “programmed” at 1 kHz compared to the cell programmed at 70 kHz.

 

Fig. 4. Normalized (ground state; $U=0\mathrm{V}$) angular distribution of scattered light power for the S-PS-DF-NLC cells programmed in the presence of an electric field of 110 V at 1 kHz (dashed curve) and at 70 kHz (solid curve).

Download Full Size | PDF

We then performed electro-optic measurements of ballistic transmission of light at normal incidence on the cell (the couple of the diaphragm and photodetector facing the incident probe beam). However, we have already presented the electro-optic characterization results of “simple” S-PSLC cells (using standard NLCs) that were programmed thanks to the positive dielectric torque (with $\mathrm{\Delta }\epsilon >0$) by using electric voltages at 1 kHz [14–16]. That is why we shall describe here the behavior of cells programmed thanks to the negative dielectric torque. Thus, Fig. 5 shows an example of the electro-optic control of light transmission in a cell that was programmed by using an electric voltage $U=110\mathrm{V}$ at 70 kHz (where $\mathrm{\Delta }\epsilon <0$).

 

Fig. 5. Stationary dependence of ballistic light transmission upon the voltage applied to the S-PS-DF-NLC cell with frequency (a) 1 kHz and (b) 70 kHz. The cell was programmed with $U=110\mathrm{V}$ at 70 kHz.

Download Full Size | PDF

As one can see [Fig. 5(a)], light transmission is relatively low at ground state (when $U=0\mathrm{V}$) and then increases with the increase of the excitation voltage at 1 kHz. The growth is not monotonic. We observe an initial increase, followed by a decrease and a final increase of transmission. As we can see also [Figs. 5(a) and 5(b)], there is some weak transmitted signal in the ground state. The application of voltages at 70 kHz reduces the transmitted signal [Fig. 5(b)] further improving the contrast of scattering modulation. Note that, here also, we see some initial oscillations, which, however, have different character (higher “frequency” and smaller modulation depth).

It is well known that the traditional scattering based devices (such as PDLCs [11]) have significant angular dependence in the transparent state (haze). We have performed haze measurements by using the setup of Fig. 2. In contrast with previous experiments all elements of the setup were kept immobile while the cell was rotated around its vertical axes for three key states of the cell. We can see (Fig. 6) that the light transmission drops below 50% of the value of normal incidence at approximately $\pm 30°$ of probe beam’s incidence angle beyond which the Fresnel reflection losses contribute significantly in the further drop of transmission.

 

Fig. 6. Angular dependence of the probe beam’s transmission of the S-PS-DF-NLC cell that was programmed with $U=110\mathrm{V}$ at 70 kHz. Squares, circles, and triangles represent, respectively, ground state ($U=0\mathrm{V}$) as well as excited homeotropic (110 V at 1 kHz) and excited planar (60 V at 70 kHz) states.

Download Full Size | PDF

5. Dynamic Characterization

Dynamic electro-optic measurements were then conducted to analyze the transitions between different states of the cell. First, the transmission was measured (Fig. 7) when a low-frequency excitation (110 V at 1 kHz) was applied to the cell to bring it to the excited homeotropic (transparent) state. Then the transmitted signal was monitored during the natural relaxation of director’s orientation, after switching off the excitation voltage ($U=0\mathrm{V}$ at $t\approx 5.6\mathrm{s}$). Finally, the same voltage (110 V at 1 kHz) was re-established (at $t\approx 12.4\mathrm{s}$) to transfer the cell from the ground state back to the homeotropic excited state. As expected, this last transition is very fast (being “forced”), lasting less than 10 ms. The natural relaxation process being particularly interesting, we present (in Fig. 8) a “zoom” of the initial stages of this process (from Fig. 7). As we can see, the relaxation process is nonmonotonic and relatively long ($\approx 0.5\mathrm{s}$). We can also see (Fig. 8) that the established (in the ground state) transmission values are still noticeably high while differing for vertical and horizontal polarizations of the incident light. Thus, the contrast of transition (or modulation depth) between the transparent (excited homeotropic) and scattering (ground) states is not very high and, in addition, it is different for two polarizations. An alternative type of dynamic transition was studied (Fig. 9) by initially applying a low-frequency voltage (110 V at 1 kHz), but the excitation signal was not switched off. Instead, the voltage and frequency of excitation were directly changed to 80 V at 80 kHz (at $t\approx 4.4\mathrm{s}$). Thus, the cell was transferred directly from the excited homeotropic state to the excited planar state. Finally, the initial excitation signal (110 V at 1 kHz) was re-established at $t\approx 11.2\mathrm{s}$.

 

Fig. 7. Transmission versus time for different transitions for two perpendicular polarizations (solid line, vertical; dashed line, horizontal). Excitation switching is performed (at $t\approx 5.6\mathrm{s}$) from 110 V at 1 kHz to 0 V and then (at $t\approx 12.4\mathrm{s}$) from zero to 110 V at 1 kHz. The S-PS-DF-NLC cell was programmed with 110 V at 70 kHz.

Download Full Size | PDF

 

Fig. 8. Transmission versus time during the natural relaxation process for two perpendicular polarizations (solid line, vertical; dashed line, horizontal). The S-PS-DF-NLC cell was programmed with 110 V at 70 kHz.

Download Full Size | PDF

 

Fig. 9. Transmission versus time for an S-PS-DF-NLC cell that was programmed at 70 kHz. Switching is performed at $t\approx 4.4\mathrm{s}$ from 1 kHz (110 V) to 80 kHz (80 V) and back to 1 kHz (110 V) at $t\approx 11.2\mathrm{s}$. Curves for two polarizations (solid line, vertical; dashed line, horizontal) are practically coinciding.

Download Full Size | PDF

We can see (Fig. 9) the following specificities in the case of switching the control frequency from 1 to 80 kHz: First, the transition from the homeotropic excitation state (110 V at 1 kHz) to the planar excitation state (80 V at 80 kHz) is monotonic and noticeably faster ($\le 10\mathrm{ms}$) compared to the free relaxation ($\approx 0.5\mathrm{s}$). Second, the polarization dependence is now practically eliminated. Third, the transmission in the planar-excited state is very low. Thus, the modulation contrast $M$ is significantly higher. However, we can also see (Fig. 9) that the back transition (from excited planar to excited homeotropic state) is now noticeably slower ($\approx 0.3\mathrm{s}$) compared to the transition from the ground state to the excited homeotropic state ($\le 10\mathrm{ms}$, Fig. 7).

6. Discussion

The electro-optic and photopolymerization procedure (programming) used in our work has provided cells with strong light scattering in their ground state ($U=0\mathrm{V}$) in contrast to the case of volume (or bulk) dispersed DF-NLC material systems, reported in [13], which were transparent at $U=0\mathrm{V}$. Those bulk composites became scattering when high-frequency voltage was applied to the cell. In our case also the cell is scattering when being excited by high-frequency electric signal. However, the application of low-frequency electric voltage brings our cell into the transparent state. There are other features in our material system (for example, the nonmonotonic behavior), which require further morphological characterization to better understand their origin. For the moment, we could speculate that the nonmonotonic behavior of transmission [upon voltage, Fig. 5(a)] could be related to the coherent quenching phenomena since we are using a coherent probe beam and we capture the transmitted signal (through the diaphragm) in a rather small acceptance angle [18,19]. The different character of nonmonotonic decrease (higher “frequency” and smaller modulation depth) in the case of using negative torque [Fig. 5(b)] could be related to smaller director clusters and higher granularity (more grains with smaller sizes) of the scattered speckle pattern.

It is interesting to analyze the observed dynamic changes that were introduced by the use of DF-NLC in the surface polymer stabilized cells (instead of simple NLCs [14–16]. First of all, we were expecting that the use of DF-NLC would allow us to obtain very fast transitions. Indeed, the transition from excited homeotropic to excited planar states is significantly faster compared to the free relaxation. For comparison purposes, we can use the typical value of $\gamma \approx 300\mathrm{mPa}\mathrm{s}$ [20] and $K\approx 15.8\mathrm{pN}$ [21] of the used DF-NLC mixture to estimate the characteristic free relaxation time ${\tau }_{\text{off}}\approx 4.8\mathrm{s}$ of the standard (pure) DF-NLC layer of thickness $d=50\mathrm{\mu m}$. As we can see from Fig. 8, the light transmission is first quickly reduced (within $\le 20\mathrm{ms}$) followed by a slow stabilization of $\approx 0.5\mathrm{s}$. We believe that the fast transition might be related to the presence of small-scale director orientation defects, predominantly near to the cell surfaces. This is confirmed by our preliminary morphological studies (Fig. 10). Indeed, we can see that the typical nonuniformities of the director orientation defects range from $<5\mathrm{\mu m}$ up to 50 μm. Note that the slower relaxation mode ($\approx 0.5\mathrm{s}$) is still significantly faster (almost by an order of magnitude) compared with the pure NLC’s relaxation ($\approx 4.8\mathrm{s}$).

 

Fig. 10. Ground-state microphotography (by using Zeiss polarization microscope) of the S-PS-DF-NLC cell that was programmed at 70 kHz.

Download Full Size | PDF

We can also see (from Figs. 7 and 8) that there is still some polarization dependence (despite the fact that the PI surfaces were not rubbed) and the contrast $M$ of light transmission’s modulation between the transparent (excited homeotropic) and ground state (via free relaxation) is not very high, ranging (for two polarizations) from $M=10$ to $M=58$. We believe that this polarization dependence may be related to the fact that the initial drop of NLC was mechanically dispersed (by us) on the cell surface to cover the entire working zone.

We can finally see (from Fig. 7) that the transition from the ground state to the excited homeotropic state (at approximately $t=12.4\mathrm{s}$) is rather fast ($\le 10\mathrm{ms}$), which is not surprising since, as we have already mentioned, the typical excitation of the NLC’s forced reorientation may be reduced by applying stronger excitations voltages [2,8].

Most importantly, we can see from Fig. 9 that the above-mentioned drawbacks (slow relaxation, polarization dependence, and low contrast) are eliminated if we use “forced relaxation” (see below) thanks to the DF-NLC. Indeed, in this case, the transition time between the transparent (excited homeotropic) and scattering (excited planar) states is much faster (milliseconds). In addition, the modulation depth is significantly improved ($M\approx 173$) and the polarization dependence is eliminated.

However, we now face another problem, which is the back transition to the transparent (excited homeotropic state). Unfortunately, it appears (Fig. 9) that this transition now is slowed significantly. We think that this could be related to the strong repulsion of the director during the negative torque (excited planar state), which could eventually force the polymer aggregates of the surface to participate in the reorientation process as it is schematically demonstrated in Fig. 11. Thus, Fig. 11(a) shows the freely relaxed NLC molecules (dotted ellipses) in their ground state after being reoriented by an electric field of low frequency (via the positive torque) to be perpendicular (solid ellipses) to the cell substrate’s surface AB. The solid bold curve 1 shows an angularly “mobile” part of a polymer aggregate, which is “attached” to the substrate on the left side of the image (the zone that is delimited by the vertical dashed line C). Figure 11(b) shows the same case of reoriented molecules (solid ellipses) as well as the case when the NLC molecules are “pushed” away (dotted ellipses) from the electric field with high frequency (via the negative torque) to be parallel to the surface AB. The bold dashed curve 2 [in Fig. 11(b)] shows the resulting possible reorientation of the polymer aggregate due to negative torque and the anchoring of NLC molecules to that aggregate.

 

Fig. 11. Schematic demonstration of the possible mechanism of slowing of the S-PS-DF-NLC’s reaction due to the difference between (a) free relaxation and (b) forced back reorientation. The polymer aggregate (solid curve 1) is bent (dashed curve 2) toward the surface AB by the negative torque.

Download Full Size | PDF

The reorientation of molecules toward the normal of the surface AB (by the positive torque) may thus be relatively faster, for the case described in Fig. 11(a), compared to the case of Fig. 11(b) since, in the second case, the polymer aggregate would also participate by adding an additional “slow step” in the reorientation process.

In conclusion, we believe that the use of DF-NLC in the S-PSLC geometry may bring noticeable benefits. First of all, it allows various modes of cell programming (negative or positive torques). Second, as we saw in this work, it may greatly improve the contrast and polarization independence of light scattering. In addition, we achieved very short transition times when switching the low-frequency excitation to high frequencies (at the order of 1 ms, in the best case). However, its promise of significantly shortening the cycle of on–off–on transition was not demonstrated, very likely because of the specific surface polymer morphology. At this stage, we do not know yet how to improve the “slowed” transition without changing the surface of the cell and without allowing further interpenetration of the RM and NLC molecules (which would bring the material system closer to the PSLC case, [13]). However, we believe that, if needed, the total on-off-on cycle duration may be further optimized by the appropriate choice of the excitation amplitudes at higher frequencies (via the negative torque). Also, another potential drawback must be taken into account (when considering the use of DF-NLCs) that is the strong temperature dependence of the excitation frequency zones for negative and positive torques [20,21].