**Dye-Doped ZnO Microcapsules for High Throughput and Sensitive Optofluidic Micro-Thermometry**

#### **Najla Ghifari 1,2,†, Sara Rassouk 1,†, Zain Hayat 1,‡, Abdelhafed Taleb 3, Adil Chahboun <sup>2</sup> and Abdel I. El Abed 1,\***


Received: 7 December 2019; Accepted: 7 January 2020; Published: 17 January 2020

**Abstract:** The main objective of this work is to show the proof of concept of a new optofluidic method for high throughput fluorescence-based thermometry, which enables the measure of temperature inside optofluidic microsystems at the millisecond (ms) time scale (high throughput). We used droplet microfluidics to produce highly monodisperse microspheres from dispersed zinc oxide (ZnO) nanocrystals and doped them with rhodamine B (RhB) or/and rhodamine 6G (Rh6G). The fluorescence intensities of these two dyes are known to depend linearly on temperature but in two opposite manner. Their mixture enables for the construction of reference probe whose fluorescence does not depend practically on temperature. The use of zinc oxide microparticles as temperature probes in microfluidic channels has two main advantages: (i) avoid the diffusion and the adsorption of the dyes inside the walls of the microfluidic channels and (ii) enhance dissipation of the heat generated by the focused incident laser beam thanks to the high thermal conductivity of this material. Our results show that the fluorescence intensity of RhB decreases linearly with increasing temperature at a rate of about −2.2%/◦C, in a very good agreement with the literature. In contrast, we observed for the first time a nonlinear change of the fluorescence intensity of Rh6G in ZnO microparticles with a minimum intensity at a temperature equal to 40 ◦C. This behaviour is reproducible and was observed only with ZnO microparticles doped with Rh6G.

**Keywords:** micro-thermometry; laser induced fluorescence; droplet microfluidics; zinc oxide; rhodamine B; rhodamine 6G

#### **1. Introduction**

With the advent of microfluidics and the development of many lab on a chip (LOC) applications, the need to ensure a good control and monitoring of temperature at the microscale with a high spatial and temporal accuracy started to merge. For instance, in biology and medicine, droplet microfluidics based on polymerase chain reaction, or digital polymerase chain reaction (dPCR), proved to be a highly sensitive technique for gene detection and gene sequencing [1,2]. This powerful technique relies on three major steps, which are performed at three different temperatures (95 ◦C, 54 ◦C and 72 ◦C) and repeated for 30 or 40 cycles to amplify a diluted template deoxyribonucleic acid (DNA) sample

enabling to reach a detectable amount of suitable fluorogenic probes. Nevertheless, despite some progress [3], the realization of an integrated microfluidic digital PCR microchip, where the whole PCR thermo-cycling process can be realized, is still hampered by the difficulty to measure and to control precisely the temperature inside the microfluidic channels.

During the last two decades, several methods, mainly based on micro-electro-mechanical systems (MEMS) technology, have been developed to ensure a good thermal control within micro-systems and micro-devices [4]. Nevertheless, most of these methods suffer from several limitations such as high cost, complicated manufacturing and poor spatial resolution. Laser induced fluorescence (LIF) technique has proven to circumvent many of these limitations [3–14]. Indeed, many fluorescent dyes exhibit a fluorescence intensity which depends strongly on the temperature. Therefore, such dyes may be used as molecular probes to measure temperature in a highly sensitive and very localized manner, both in space and time. Among the most efficient molecular probes for temperature measurements, rhodamine B (RhB) is the most widely used.

At low concentrations, the fluorescence intensity *If* (per volume unit) of a fluorescent dye depends generally on several parameters according to the following linear relation, which may be derived simply from Beer-Lambert law [15–17]:

$$I\_f \propto \Phi I\_0 \mathbf{e} \mathbf{C} \tag{1}$$

where Φ, *I*0,  and *C* are the fluorescence quantum yield, incident illumination intensity, molar absorptivity and dye concentration, respectively. For most temperature sensitive dyes, likewise RhB, it is generally observed that an increase in temperature results in a decrease of the fluorescence quantum yield Φ and the fluorescence life time. This is because the non-radiative processes related to thermal agitation (collisions with solvent molecules, intra-molecular vibrations and rotations) are more efficient at high temperatures [15,18]. One may cite nevertheless, the particular cases of rhodamine 6G (Rh6G) whose fluorescence increases linearly with temperature [6] and rhodamine 110 (Rh110) whose fluorescence does not depend on temperature. For the latter compounds, the increase or stability of fluorescence intensity versus temperature may be attributed to the formation of complexes (dimers, trimers, ...). Such complexes are better shielded from inter-molecular collisions and have higher quantum yields leading hence to an increase or a stability of fluorescence intensity versus temperature [18]. As can be seen from the above equation (Equation (1)), factors such as a non-uniform illumination (*I*0) or a non-uniform dye concentration *C* can lead to major errors and uncertainties on temperature measurement.

While uncertainty caused by a fluctuation of the laser illumination can be eliminated relatively easily by normalizing fluorescence intensity using a reference temperature-insensitive dye, such as Rh110, those which are caused by a fluctuation of the dye concentration in the microfluidic device are much more difficult to handle. Indeed, several processes can be at the origin of concentration fluctuations, such as adsorption of dye molecules on the channel walls [7], trapping of molecules inside the porous substrate of the microfluidic device (e.g., PDMS) or a non homogeneous distribution of the flow advected dye molecules along the microchannel, caused by the well-known Taylor dispersion phenomena [19,20]. The existence of all these processes may result in a local depletion or enrichment of the dye in the microfluidic device.

One should emphasize also another potential and important source of uncertainty on the measure of temperature in microfluidic systems, that is the heat dissipation within a tiny volume illuminated by a focused laser beam. To our knowledge, such an effect has been often neglected in the literature. In order to minimize the increase of temperature caused by the heating of the incident laser beam, one should either use a highly sensitive detection setup enabling to minimize the illumination power and/or increase the efficiency of the heat dissipation in the illuminated area of the sample by using a suitable material with a high thermal conductivity.

The main objective of this work is to give a proof of concept of a new approach for measuring the temperature inside microfluidic channels at high throughput and in a reliable manner using laser induced fluorescence (LIF) technique of rhodamine B dye confined in highly monodisperse zinc oxide

microspheres. Indeed, besides its high chemical stability and photo-stability, zinc oxide exhibits a high thermal conductivity of about 100 to 150 W/m·K (depending on ZnO structural properties) [21–23]. This value is for instance two orders of magnitude higher than the thermal conductivity of water, i.e., 0.65 W/m·K. We used three types of dye doped ZnO microparticles: one type is doped with RhB, a second type is doped with Rh6G and the third type is doped with a mixture of RhB and Rh6G. The later mixture is intended to be used as a reference for fluorescence intensity calibration and to take account of fluctuations in incident illumination intensity. Indeed, the fluorescence intensity of RhB decreases linearly with increasing temperature at a ratio of about −2%/◦C whereas the fluorescence intensity of rhodamine 6G (Rh6G) increases linearly with increasing temperature at a rate of about +2%/◦C [6,12,24,25], as shown (see Figure 1).

**Figure 1.** Fluorescence spectra of Rhodamine 6G (**A**) and Rhodamine B (**B**) versus temperature.

#### **2. Experimental Section**

#### *2.1. Synthesis of ZnO Nanoparticles Building Units*

We first synthesized ZnO nanoparticles which serve as building blocks for the final ZnO microparticles using sol-gel technique. Then, we confined such nanoparticles in highly monodisperse droplets using a flow-focusing microfluidic device in order to fabricate highly monodisperse ZnO microspheres where RhB and Rh6G are confined in a controlled manner.

The synthesis of colloidal ZnO nanoparticles can be carried out using different types of precursors dissolved in alcohol solvents [26]. We used zinc acetate dehydrate Zn(CH3COO)2:2H2O (99.999%, from Sigma-Aldrich France, St. Louis, MO, USA) as ZnO precursor and methanol as a solvent, following the procedure which is described in more detail in [27]. All reagents were of analytical grade and were used as received without any further purification. Briefly, we dissolved 0.6 g in 5 mL of methanol (CH3OH), then, the solution was stirred during 1 h at 60 ◦C under magnetic stirring to ensure homogeneous mixing and obtain a transparent solution.

The mechanism of formation of zinc oxide nanoparticles used in this work is according to a succession of chemical reactions. First, the dissolution of dehydrated zinc acetate in the presence of methanol contributes to the dehydration of zinc oxide precursor which results in the formation of anhydrous zinc acetate and water. Followed by a preliminary dissolution of the anhydrous zinc acetate to form the zinc ion and acetate. The latter will become acetic acid. It then takes place a chemical reaction between the species present (zinc and hydroxide ions) in the solution causing the precipitation of zinc hydroxide, which is ultimately converted to ZnO nanoparticles. Finally, the resulting dispersion of ZnO nanoparticles (with residual solvents) is doped with the desired fluorescent dye/dyes. Subsequently, the resulting mixture was immediately injected into the microfluidic chip to generate highly monodisperse microdroplets. The formation mechanism of zinc oxide nanoparticles can be summarized in these chemical reactions:

$$\text{Zn(CH}\_3\text{COO)}\_2 \cdot 2\text{H}\_2\text{O} \longrightarrow \text{Zn(CH}\_3\text{COO)}\_2 + 2\text{H}\_2\text{O}$$

$$\text{Zn(CH}\_3\text{COO)}^{2-} \longrightarrow \text{Zn}^{2+} + 2\text{CH}\_3\text{COO}^-$$

$$\text{Zn}^{2+} + 2\text{OH}^- \longrightarrow \text{Zn(OH)}\_2$$

$$\text{Zn(OH)}\_2 \longrightarrow \text{ZnO} + \text{H}\_2\text{O}$$

#### *2.2. Fabrication of Microfluidic Devices and Synthesis of ZnO Microparticles*

Microfluidic devices were manufactured according to the conventional soft lithography technique [28]. In a first step, a pattern was transferred to SU-8 photoresist, previously coated on a silicon wafer, followed by the exposure to ultraviolet (UV) light, through the mask pattern. UV illumination leads to the polymerization of the photoresist located under the transparent regions of the mask. After development, the master mold is ready for the next step. In a second step, the polydimethylsiloxane (PDMS) was first mixed with a cross-linking agent with a weight ratio of 10:1, the mixture then was degassed using a vacuum pump at room temperature and the solution was poured onto the previously fabricated mold and placed in the oven for polymerization at 75 ◦C for 2 h. The block of PDMS was then removed from the mold; we thus obtain a replica of microchannels. In a third step, the PDMS block and the glass slide were treated with oxygen plasma for 20 s to enable their bonding and sealing the microfluidic chip. The design of the microfluidic device for droplets generation contained two inputs, one input for the carrier oil and a second input for the dispersed phase, a main (square) channel cross section of about 80 μm × 80 μm and an output for the collection of droplets in a Petri dish (see Figure 2a,b). The flow rates of the carrier oil (*Qc*) and the dispersed phase (ZnO dispersion) (*Qd*) were set using Nemesys syringe pumps (Cetoni GmbH, Korbussen, Germany). The dispersed phase consisted of ZnO solution doped with one of the selected rhodamine dyes: RhB, Rh6G and a mixture of both dyes. The flow rates were set in order to produce droplets with the same size (55 μm) for all experiments. Then, the droplets were collected in Petri-dish with HFE 7500 at room temperature.

The used carrier oil phase consisted of a fluorocarbon oil (HFE 7500, 3-ethoxy-dodecafluoro-2 trifluoromethyl-hexane, Inventec, Bry-sur-Marne, France), with a density of 1.61 g/cm3 and thermal diffusivity of about *<sup>κ</sup>* = 3.6 × <sup>10</sup>−<sup>8</sup> m2/s; 0.2 % (*w*/*w*) of a commercial surfactant (dSURF, Fluigent, Le Kremlin-Bicêtre, France) was added in HFE 7500 oil in order to prevent droplets merging. The used fluorocarbon oil has the advantage of not inducing PDMS swelling and being also chemically inert. More important for our application, it does not solubilize any of the ZnO precursor solution components.

After their generation, droplets were transported along the microfluidic channel by the carrier oil phase (see Figure 2) and then collected in a Petri dish (not shown on Figure 2), where droplets formed a floating monolayer at the oil-air interface, because of the higher density of fluorocarbon oil (*dHFE*-7500 = 1.62). In our study, the condensation of ZnO droplets was mainly controlled by the evaporation of the solvent at the oil–air interface. We let the condensation performing until microparticles reach their final sizes: 17.8 ± 0.2 μm for RhB doped microparticles and 20 ± 0.2 μm for both Rh6G and RhB/Rh6G doped microparticles.

For the re-injection of the synthesized microparticles, we used another microfluidic device (re-injector), which was made of a main inlet enabling the injection of the dispersed ZnO microparticles in HFE 7500 oil and three other secondary inlets for the injection of pure carrier oil to prevent aggregation of microparticles and the clogging of the microfluidic channel. The design of the re-injector includes also a constriction of the rectangular channel, with a width *w* = 22 μm and a height *h* = 38 μm. The overall flow rate in this constriction area was about *Q* = 150 μL/h. Therefore a mean flow velocity of about *vconstr*. <sup>=</sup> *<sup>Q</sup> <sup>w</sup>* <sup>×</sup> *<sup>h</sup>* 5 cm/s can be deduced.

**Figure 2.** (**A**) Optical micrograph and schematic illustration of (**B**) the flow-focusing microfluidic geometry for zinc oxide droplet formation; (**C**) and the different case studies; (**D**) Optical micrograph of stable and monodisperse zinc oxide droplets generated through the flow-focusing microfluidic device, Scale bar, 100 μm; (**E**) Schematic illustration of the used microfluidic design for fluorescence analysis of doped ZnO microparticles.

#### *2.3. Fluorescence Detection of Flowing Microparticles*

Figure 3 shows the home-built fluorescence acquisition setup used in this study. It enables for a highly sensitive fluorescence detection, which is described in detail in Ref. [29]. It includes multiple laser sources optimised for the absorption of different fluorophores (we used in this study one continuous-wave (CW) source with wavelength of 532 nm). Laser incident beams are combined by mean of a first dichroic mirror (DM1) and then directed towards microdroplets in the microfluidic channel by mean of a second dichroic mirror (DM2). The focused band limited light is targeted towards the droplets/particles and recollected by the microscope objective, which is then transmitted through another set of band-limited filters to two photo-multiplier tubes (PMT's). The signal outputs from PMT's are then collected at high acquisition rates (100 KHz) using a DAQ acquisition card (National Instruments) and analyzed them using LabVIEW and a FPGA (field programmable gate arrays) module scripts, which allows for the identification of droplets by the modulation of fluorescence versus time. Depending on the selected sampling acquisition of the fluorescence signal, the width of the time lapse for signal detection can be varied from 40 ms (max sample rate of 200 KHz) to seconds or minutes.

**Figure 3.** Experimental setup for high throughput fluorescence measurements.

#### *2.4. Calibration of Temperature Measurements*

We used a thermo-plate heating chamber (Tokai-Hit, Shizuoka, Japan) to control precisely the temperature of the whole microfluidic device (±0.1 ◦C). This device is equipped with two heating plates, one at the bottom and one at the top. The later consists of a large clear glass top heater providing a uniform temperature distribution in the whole chamber. The heating device is also equipped with a feedback sensor mechanism that enables a real-time, precise sample temperature feedback temperature regulation. The heating chamber was placed on IX73 Olympus microscope (OLYMPUS, Tokyo, Japan) stage to measure the fluorescence of ZnO particles doped with rhodamine at different temperatures from 20 to 50 ◦C.

To determine the dependence of the fluorescence intensity versus temperature, we enclosed the entire microfluidic chip in the heating chamber. For each temperature value, we waited at least 10 min before recording the fluorescence intensity. Also, in order to give enough time for the incoming microparticles and HFE 7500 oil to reach the selected (target) temperature, a length of the tubing inlets (ZnO microparticles dispersion and pure fluorocarbon oil) of about *L* 10 cm was enclosed in the heating chamber. Taking account of the used overall flow rate, *Q* 150 μL/h and the inner diameter of the tubing, *Dtub*. = 0.56 mm, one may deduce a flow velocity in the tubing of about *vflow* <sup>=</sup> *<sup>Q</sup> <sup>π</sup>*(*Dtub*./2)<sup>2</sup> 4 cm/min.

This means that it should take less than 3 min for the incoming fluid and microparticles to flow from the entry of the heating system until the inlet of the microfluidic channel, where fluorescence intensity is measured.

Considering now the thermal diffusivity of HFE 7500 oil, that is 2.16 mm2/min, and the cross section size of the tubing, one may calculate a diffusion time as small as *tdi f f* 4 s, which is the time needed for the temperature to diffuse and become homogenized through all the cross section of the flowing fluid. As one may notice, the calculated diffusion time is much shorter than the flow time of the fluid before it reaches the area of the microfluidic channel where the temperature is measured.

#### **3. Results and Discussion**

#### *3.1. Fluorescence Intensity Versus Temperature of ZnO Microparticles Doped with rhodamine B (RhB)*

Figure 4 shows the change versus different temperatures, ranging from 20 to 50 ◦C, of the fluorescence intensity of RhB doped ZnO microparticles carried by a flow of HFE 7500 of about 150 μL/h in the microfluidic channel.

We first show in Figure 4A the fluorescence intensity peaks of individual flowing microparticles as they were detected and recorded from different experiments (at different temperatures). Because microparticles were randomly dispersed in the carrier oil, the time at which microparticles crossed the constriction area of the microfluidic channel, where they were detected, was random and hence the origin of the time axis was arbitrary. Nevertheless, one could have the measured value of the width of each peak to roughly estimate the size of the microspheres, which was found here to be about 1.5 ms, as shown in Figure 4B.

**Figure 4.** (**A**) Recorded fluorescence intensity peaks at different temperature from confined rhodamine B dye in flowing ZnO microcapsules (with a size of 17.8 μm) along a 22 μm wide micro-channel constriction area; (**B**) High-magnification of the fluorescence intensity peak vs. time of doped ZnO microcapsules with rhodamine B; (**C**) Plot of the linear decrease of the normalized fluorescence intensity of doped ZnO microparticles with rhodamine B vs. temperature; (**D**) Comparison of the change of fluorescence of RhB vs. temperature in flowing RhB doped ZnO microparticles (squares, red curve) and flowing RhB doped ZnO microdroplets (circles, blue curve), the slope of the two curves are very close and in very good agreement with the literature data: (−2.2 ± 0.1)%/◦C and (−1.9 ± 0.2)%/◦C, respectively.

Taking account the mean flow velocity of 5 cm/s in the constriction area of the microchannel, one deduces a microparticle size of about *L* 75 μm, which was four times the size we measured from microscopy image analysis, *DRhB* = 17.8 ± 0.2 μm. Noteworthy, we made the assumption that microparticles flow with the same mean velocity as the surrounding carrier fluid's, which was basically not a good approximation, because of the presence of a thin lubrication film of the continuous phase between the particles and the channel walls as reported in the literature [30,31].

It is also interesting to notice from Figure 4B that the fluorescence peaks exhibited an asymmetric bell shape with an elongated tail at the back side of the microparticle. This observation, with other results obtained from optical and scanning electronic microscopies (not shown in this study), corroborates the relative softness of the synthesized microparticle and their deformability. We have shown in a previous study that the synthesized microparticles consist of microcapsules with a hollow structure and a thin envelope whose thickness value was found to be about 0.7 μm [27]. The group of Slasac and Barhès-Biesel et al. have shown for instance that microcapsules can deform easily under the effect of a shear flow in a microchannel [32,33].

One of the main advantage of droplet microfluidics based synthesis approach is the ability to obtain highly monodisperse microparticles, which in turn enabled us to obtain results with very good statistics. We normalized the fluorescence intensity values by considering the intensity recorded at 20 ◦C as a reference, *I*20◦*<sup>C</sup>* = 1.

For each temperature, we collected the fluorescence signal from approximately 20 highly monodisperse microparticles and plot the average values with their standard deviation versus temperature, as shown in Figure 4C. This figure shows clearly that the fluorescence intensity of RhB doped ZnO microparticles decreased linearly vs. temperature with a rate of about (−2.2 ± 0.1)%/◦*C*. The determined slope was in very good agreement with the value reported generally in the literature for RhB dye [4–6,12–14].

In order to compare the effect of ZnO on the fluorescence behavior of RhB versus temperature, we present in Figure 4D the (normalized) fluorescence intensity of RhB when confined in microdroplets made of a water based RhB solution (0.5 mM and droplets size about 50 μm). We found that the slope of the linear decrease of the fluorescence intensity versus temperature was very close to the one recorded from flowing RhB doped ZnO microparticles, (−1.9 ± 0.2)%/◦C.

#### *3.2. Fluorescence Intensity Versus Temperature of ZnO Microparticles Doped with rhodamine 6G (Rh6G)*

Figure 5A shows the change versus temperature of the fluorescence intensity of ZnO microparticles when doped with Rh6G dye. As may be noticed, though the fluorescence intensity appeared to increase globally with temperature, as observed in the bulk solution [6], we observed for the first time that the fluorescence intensity exhibited a minimum around 40 ◦C. This behavior is reproducible and was observed only when ZnO microparticles were doped with Rh6G.

Figure 5B shows a "standard" linear increase versus temperature of the of the fluorescence intensity of Rh6G when dissolved in microdroplets made of a 0.5 mM aqueous solution (droplet size equal to 50 μm) and when dissolved in a bulk solution. Nevertheless, the obtained slope for both curves, around 0.5%/◦C, was much smaller than the one reported in the literature, 1.9%/◦C [6]. Nevertheless, no minimum of intensity was observed in these cases, which indicates that this phenomena must be related to the confinement of Rh6G in ZnO microparticles.

**Figure 5.** Change of the relative intensity of the fluorescence of rhodamine 6G (Rh6G) vs. temperature in (**A**) ZnO microparticles and (**B**) in water droplet (blue curve) and in ethanol droplet (red curve).

Though, the aim of this study is to give a simple demonstration of the proof of concept of our dye doped ZnO microparticles approach for locally measuring the temperature, we present preliminary results obtained by optical and scanning electron microscopies, which enables to understand (partly) the observed difference between RhB and Rh6G dyes in ZnO microparticles, as shown in Figure 6.

**Figure 6.** Optical microscopy images of doped ZnO microspheres with RhB (**A**), Rh6G (**B**) and a RhB/Rh6G mixture (**D**); (**C**) High resolution scanning electron microscopy image of a ZnO microparticle showing the aggregation of ZnO nanocrystal building units of the microsphere and the resulting porosity of the material.

Figure 6A shows that RhB molecules organized in a homogeneously distributed pink colored shell around ZnO microsphere, which corresponded basically, as stated earlier, to a microcapsule with a thin porous envelope having a thickness of about 0.7 μm, as shown in more detail in reference [27]. We emphasize in this study the porosity of the microcapsule shell, which resulted from the aggregation of ZnO nanoparticles building units, whose average size was found to be about 100 nm, and the formation of voids between such nanoparticles, as can be seen from the scanning electron microscope (SEM) image shown on Figure 6C. One may suggest also that these pores may serve as niches for aggregates of dyes molecules.

Also, since rhodamine molecules were added at the early stage of the formation of the microcapsules, i.e., in the microdroplet state, the observed colored ring on the surface of ZnO microsphere should correspond to an adsorbed layer of RhB molecules on the inner surface of the microcapsule. In contrast, Figure 6B shows that Rh6G dye molecules were less localized on the surface of the microcapsule and the dye seemed to occupy a diffuse and extended area all over the surface of the shell. Moreover, we observe the presence of many bright spots with different brightness, which revealed a non homogeneous distribution of Rh6G dye molecules inside the shell of the microcapsule. We attribute such brighter regions to pores and niches in the ZnO microcapsule shell where the dye accumulates more or less strongly. Moreover, though we have used the same size for the initial droplets (55 μm) and the same ZnO precursor concentration (0.25 mM) for both types of microcapsules, we found that those doped with Rh6G had a lager size (20 ± 0.2 μm) than those doped with RhB (17.8 ± 0.2 μm).

We suggest that the two different behaviors of the two dyes with ZnO microcapsules should be related to the electrical charges carried by the dyes and the electrically charged ZnO nanocrystals building blocks. Indeed, whereas RhB molecules can adopt (depending on the pH value) either a positively charged form or a zwitterionic form, for which the overall electrical charge is zero, Rh6G molecules can carry only a positive electric charge, as shown from their chemical structures in Figure 7A. Also, ZnO nanocrystals are known to carry a net electric dipole moment, which is inherent to the tetrahedral configuration of the wurtzite crystalline structure. In a such structure, each type of ion, Zn2<sup>+</sup> or O2−, has four neighbouring ions of the other type of atom, and vice versa. That is to say, the tetrahedral coordination exhibits a sequence of positively charged Zn2<sup>+</sup> and negatively charged O2<sup>−</sup> polar planes within the c-axis direction, thus respectively contributing to two opposite faces of polarity (0001) and (0001¯) perpendicular to the c-axis. This results in an intrinsic electric dipole moment and a spontaneous polarization along the c-axis of the nanocrystals [34–38].

Therefore, one may expect that a strong interaction should occur between the positively charged Rh6G molecules and the negatively charged plane of ZnO nanocrystals, leading probably to a larger diffusion of Rh6G inside the porous structure of the microcapsule shell than could occur with RhB molecules in their zwitterionic form. At this stage, we still need further investigations in order to make a correlation between the adsorption of Rh6G dye in the pores of the shell and the observation of a minimum fluorescence intensity at a temperature around 40 ◦C.

**Figure 7.** (**left**) Chemical structure of the used RhB and Rh6G dyes, depending on the pH, RhB molecules may exhibit either a zwitterionic form (presence of both a positive and negative charges are present) or may exhibit only a negative charge, whereas Rh6G molecules can carry only a positive charge or no charge at all, (**right**) illustration of the formation of the polar ZnO nanocrystals building units.

#### *3.3. Fluorescence Properties of ZnO Microparticles Doped with a Mixture of RhB and Rh6G*

Figure 8 shows the fluorescence intensity versus temperature of ZnO microparticles which were doped with a mixture of RhB (29 μm) and Rh6G (37 μm) dyes. Because of the non linear variation of the fluorescence intensity of Rh6G in ZnO microparticles, it was difficult to predict the concentration ratio of the two dyes which would lead to a constant fluorescence intensity versus temperature. Nevertheless, after several attempts, we found that the ratio [*RhB*] [*Rh*6*G*] <sup>=</sup> <sup>29</sup> <sup>μ</sup>*<sup>M</sup>* <sup>37</sup> <sup>μ</sup>*<sup>M</sup>* gave a good reference for which the fluorescence intensity remained practically constant versus temperature, as shown in Figure 8.

**Figure 8.** Relative intensity of the fluorescence of RhB and Rh6G mixture versus temperature in ZnO microparticles.

#### **4. Conclusions and Perspectives**

We have shown in the present study that ZnO microparticles doped with RhB dye exhibit a fluorescence intensity which decreases linearly versus temperature with a rate of about (−2.2 ± 0.1)%/◦C. This value is in very good agreement with the literature date. Our results should pave the way for the development of a promising technique which will enable in the future for a real time and highly sensitive measurement of the temperature in microfluidic channels. The use of ZnO microparticles for confining the temperature sensitive dyes has two main advantages: (i) it enables to circumvent the diffusion and the adsorption of the dyes on the walls of the microfluidic channel and inside the pores of the PDMS substrate of the microfluidic device, (ii) it enhances the dissipation of the heat generated by the focused incident laser beam thanks to the high thermal conductivity of ZnO material. In contrast with the fluorescence intensity linear change versus temperature of Rh6G in bulk solution or in microdroplets, Rh6G when confined in ZnO microparticles shows a non linear behavior of the fluorescence intensity versus temperature. The presence of a minimum at a temperature around 40 ◦C is reported for the first time. Such non linear behavior makes the construction of reliable reference a little bit cumbersome by nonetheless we determined experimentally a suitable mixture of RhB and Rh6G for which the fluorescence intensity remains practically constant in the investigated temperature range [20 ◦C–50 ◦C]. We envisage to consider in a future study the use of rhodamine 110 for the construction of such reference. Also, we plan to flow and detect simultaneously the two types of microparticles (RhB as a measure probe and Rh110 as reference probe) in the microchannel in order to make the normalization of the fluorescence intensity signal straightforward. For this purpose, the use of two different concentrations for the two types of microparticles will allow for the discrimination between the two types of particles while flowing at high throughput in the microchannel, as illustrated in Figure 9.

**Figure 9.** Different microparticles doped with two different dyes may be discriminated by the value of their fluorescence intensity signals. Such concentration based coding method will be applied in the future for the simultaneous recording of the fluorescence intensity signals from RhB doped microparticles (measure probes) and Rh110 doped microparticles (reference probes).

**Author Contributions:** Conceptualization, A.I.E.A.; methodology, N.G., S.R. and A.T.; validation, A.I.E.A. and A.C.; formal analysis, N.G., S.R. and A.I.E.A.; investigation, N.G. and S.R.; resources, Z.H., A.C. and A.I.E.A.; writing—original draft preparation, N.G., S.R. and A.I.E.A.; writing–review and editing, N.G., Z.H., A.T., A.C. and A.I.E.A.; visualization, Z.H.; supervision, A.C. and A.I.E.A.; project administration, A.C. and A.I.E.A.; funding acquisition, A.C. and A.I.E.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by PHC-Toubkal 2017 program (Programme Hubert Curien).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Synthesis and a Photo-Stability Study of Organic Dyes for Electro-Fluidic Display**

**Yong Deng 1,2,3, Shi Li 1, Dechao Ye 2,3, Hongwei Jiang 1, Biao Tang 1,\* and Guofu Zhou 1,2,3**


Received: 9 December 2019; Accepted: 7 January 2020; Published: 11 January 2020

**Abstract:** Electro-fluidic display (EFD) is one of the most promising reflective displays because of its full color and video speed. Colored EFD oil, which normally consists of soluble organic dyes and non-polar solvent, plays a critical role in color, electro-optical behavior, and the reliability of the EFD devices. In this paper, we report our research on two kinds of electro-fluidic dyes based on anthraquinone and azo pyrazolone, including their synthesis, structure characterization, and application properties. Changes of absorbance curves, color coordinates of oils, and photoelectric responses of devices were studied in detail under accelerated irradiation to investigate the photo-stability and reliability properties of synthesized oil materials and devices. Photoelectric responses and photo stability of dyes are highly varied depending on their structures. We found that 1,4-dlialkylamino anthraqinone and mono azo pyrazolone dyes are much more stable than 1,8-dlialkylamino anthraqinone and corresponding bisazo pyrazolone dyes.

**Keywords:** electro-fluidic display; organic dye; colored oil; photo-stability

#### **1. Introduction**

Recently, electro-fluidic display as an emerging display technology has received widespread attention [1–4]. It has many advantages: (1) a reflective mode for using ambient light and energy saving [5]; (2) quick response (its switching time is less than 20 ms) for video display [6]; (3) superb optical performance (its white state reflectance is up to 50% [7] and it has full color) [8]; and (4) fluidic and soft display candidate materials for flexible displays.

Colored oils act as optical switches in electro-fluidic display, which affect not only the color gamut but also the contrast ratio, electro-optical response, and the lifetime for outdoor display devices. High optical density, good solubility in non-polar solvent, and good light stability are the main characteristics for the designation of soluble organic dyes. A series of efforts to develop efficient dyes for electro-fluidic display have been conducted in previous research, such as the designation of anthraquinone dyes [9,10], azo dyes [11–15], dipyrrole methane metal dyes [16], and pigment dispersion [17,18]. However, for many of the dyes that were disclosed in patents, few were subjected to detailed research on their synthesis and specific application properties, especially their photo-stability properties.

In our opinion, the photo-stability of the oil materials is one of color dyes' most crucial properties, which could limit their industrialization. In their designation lifetime, color dyes will decompose and their color image as well as their response properties will deteriorate. In this context, we have synthesized two typical electro-fluidic display (EFD) dyes (anthraquinone and azo pyrazolone dyes), and their electro-optical behavior, photo-stability properties, and structure correlation were researched in detail in this paper.

#### **2. Experimental**

#### *2.1. Materials*

All of the color dyes were synthesized in our laboratory. The structures were characterized and proved by using several analysis methods: Infrared (IR) (PerkinElmer, Shelton, CT, USA), Nuclear magnetic resonance (1H NMR) (Bruker BioSpin GmbH, Rheinstetten, Germany), 13C NMR (Bruker BioSpin GmbH, Rheinstetten, Germany), High resolution mass spectra (HRMS/ESI) (Bruker Daltonics, Bremen, Germany)). Indium Tin Oxide (ITO) coated glass (0.7 mm thickness) with a resistance of 100 Ω·m (purchased from Shenzhen Laibao Hi-Tech Co. Ltd., Shenzhen, China) was used as the bottom and top substrates. SU-8 3005 photoresist was purchased from Microchem Corp. (Newton, MA, USA). AF Teflon® 1600X was purchased from DupontTM Co. (Wilmington, DE, USA)

#### *2.2. Analysis Methods*

UV-3300 spectrophotometer (Jinpeng Analytical Instrument Corporation, Shanghai, China) was used to record UV-vis (ultraviolet) absorption spectra of synthesized dyes. Perkin–Elmer 841 spectrometer (PerkinElmer, Shelton, CT, USA) using KBr pellets was used to record FTIR spectra. Varian AS400 (Agilent, Santa Clara, CA, USA), Bruker AVANCEIII 500 (Bruker BioSpin GmbH, Rheinstetten, Germany) and AVANCE NEO Bruker 600 (Bruker BioSpin GmbH) were used to record 1H NMR and 13C NMR spectra in deuterated solvents using tetramethylsilane (TMS) as an internal standard. Bruker Daltonics Microflex mass spectrometer (Bruker Daltonics, Bremen, Germany) was used to record the mass spectra. The thickness of the AF layer and height of the pixel wall were measured by using the Dektak (Dektak XT, BRUKER, Hamburg, Germany). A self-assembling oil filling instrument was made by our team. The photo-stability of Four dyes in EFD cell was tested in a Xenon arc lamp weather resistance test chamber (B-SUN-I), purchased from Shanghai Yiheng Instruments Co. Ltd (Shanghai, China). The simulation conditions were set according to international standard IEC 60068-2-5: Procedure B; 0.55 W/m2 (340 nm); Temperature 45 ◦C. Absorbance curves before and after a period of irradiation were tested.

#### *2.3. EFD Oil Formulation*

Certain quantities of dyes and decane were weighted in a sample cell and put in the ultrasonic instrument for 30 min. Afterwards, the solution was subjected to a filtration process using a 0.45 μm filtrator to remove suspended impurities. Then the well prepared oil was used in the next EFD device fabrication process.

#### *2.4. Electro-Fluidic Display (EFD) Device Fabrication*

The EFD device was fabricated as follows: a 8 <sup>×</sup> 6 cm2 ITO coated glass was used as a bottom substrate. A Teflon (AF Teflon® 1600X, DupontTM Co.) layer of 0.8 μm was spin-coated on top of the ITO glass as the dielectric layer and hydrophobic layer. A pixel wall structure was fabricated by standard lithography process using photoresist material (SU-8 3005). After coating a photoresist film on active fluoropolymer surface, an exposure process was used to obtain a patterned reacted resist layer. Finally, a 7.5 μm height pixel wall with a 15 μm width was developed. The square of each pixel was 150 <sup>×</sup> 150 <sup>μ</sup>m2. After that, oil with an average thickness of 4.5 <sup>μ</sup>m was filled into pixels by a self-assembly machine. The last step of the fabrication processes was to integrate another top ITO glass with a gap of 75 μm filled by a water phase.

#### **3. Dye Synthesis and Characterization**

The structure of four designed dyes were shown in the Figure 1.

Anthra-1: 2.0 g (242 g/mol, 0.0082 mol) Anthracene-1,4,9,10-tetraol was dissolved in 20 mL 2-methoxyethanol in a three-necked flask with agitation, and the solution was heated to 80 ◦C under an N2 atomosphere. Then, 4.26 g (129 g/mol, 0.033 mol) 2-ethyl-hexylamine was added once. The solution was refluxed for 24 h. After that, the reaction was exposed to air for 2 h at 50 ◦C. Then the reaction was cooled to room temperature, poured into 100 mL water, and filtrated to get 2.77 g crude products Anthra-1. The crude Anthra-1 was purified by flash chromatography on silica gel (eluting agent: petroleum ether and ethyl acetate). Characterization: cyan solid; yield: 73.1%; λ (max) = 650 nm (in decane); <sup>ε</sup> = 17930 L·mol−1·cm−1. IR (KBr, cm−1): 3061.80 (Ar–H); 2962.50 (–CH3), 2921.20 (–CH2 −), 2859.10 (–CH3); 1645.9 (–C=O); 1581.80 (Ar); 1555.0 (Ar); 1521.90 (Ar); 1459.90 (–CH–); 1379.27(–CH3); 1257.28 (C–N–C); 816.89. 1H NMR (500 MHz, CDCl3): 9.71 (s, 1H, Ar–NHCH2 −), 8.25 (m, 2H, Ar–H), 7.65 (m, 2H, Ar–H), 7.35 (d, 2H, *J* = 19.0, Ar–H), 3.22 (m, 4H, –NCH2 −), 1.73 (m, 2H, –CH–), 1.50 (m, 4H, –CH2 <sup>−</sup>), 1.26 (m, 6H, –CH2 <sup>−</sup>),1.18 (m, 6H, –CH2 <sup>−</sup>), 0.88 (m, 12H, –CH3). 13C NMR (500 MHz, CDCl3): 182.11, 146.56, 134.57, 131.84, 126.03, 123.66, 109.67, 45.99, 39.47, 31.28, 29.04, 24.53, 23.03, 14.12, 10.99. MS (MALDI-TOF) (DIF) M/Z (%): 463.20 (M + H)<sup>+</sup>. Calc. for (C30H42N2O2): 462.30.

**Figure 1.** Molecular structures of the four synthesized dyes.

Anthra-2: 2.0 g (227 g/mol, 0.0088 mol) 1,8-dichloroquinone was dissolved in 20 mL 2-ethyl-hexylamine in a three-necked flask with agitation, and the solution was heated to 150 ◦C under an N2 atomosphere for 12 h. After reaction, the solution was poured in 10% hydrochloric acid solution filtrate to remove excess 2-ethyl-hexylamine and to obtain 1.60 g of crude products Anthra-2. The crude Anthra-2 was purified by flash chromatography on silica gel (eluting agent: petroleum ether and ethyl acetate). Characterization: magenta solid; yield: 40%; λ (max) = 545 nm (in decane); ε = 13335 L·mol−1·cm<sup>−</sup>1. IR (KBr, cm−1): 3268.50 (–NH–); 3082.40 (Ar–H); 2962.50 (–CH3); 2925.30 (–CH2 −); 2863.30 (–CH3); 1658.30 (–C=O); 1623.20 (–C–N–); 1569.40 (Ar); 1531.60 (Ar); 1459.90 (–CH–); 1397.80 (–CH3); 1296.50 (C–N–C); 1211.80 (C–N–C); 1073.20; 1019.51; 837.57; 744.53. 1H NMR (500 MHz, CDCl3): 9.67 (s, 1H, Ar–NH–); 7.45–7.43 (d, 2H, *J* =1 Hz, Ar–H); 7.40–7.37 (t, 2H, *J* = 15.5 Hz, Ar–H); 6.95–6.93 (d, 2H, *J* = 1 Hz, Ar–H); 3.16–3.14 (s, 4H, *J* = 11 Hz, Ar–NH–), 1.65 (m, 2H, –CH–); 1.44 (m, 8H, –CH2–); 1.28 (m, 8H, –CH2–); 0.91-0.88 (t, 6H, *J* =15 Hz, –CH3); 0.86–0.83 (t, 6H, *J* = 15 Hz, –CH3). 13C NMR (500 MHz, CDCl3): 188.99, 184.84, 151.48, 134.22, 117.63, 114.63, 46.08, 38.97, 31.49, 28.99, 24.79, 23.06, 14.11, 11.13. MS (MALDI-TOF) (DIF) M/Z (%): 463.20 (M + H)<sup>+</sup>. Calc. for (C30H42N2O2): 462.30.

Pyrazolone-1: A quantity of 1.43 g (143 g/mol, 0.01 mol) naphthyl amine was dissolved in 20 mL of ethanol, 10 mL of water, and 3 mL (0.036 mol) of concentrated hydrochloric acid. After it had been rapidly cooled to 0–5 ◦C, a 5 mL solution containing 0.72 g (69 g/mol, 0.0105 mol) of sodium nitrite was added. Ehrlich reagent was used to detect the termination of diazotisation, and sulfamic acid was used to remove the residual nitrous acid. Afterwards, 2.94 g of the coupling component 1-(2-ethylhexyl)-3-(2-ethylhexyl)-5-pyrazalone (294 g/mol, 0.01 mol) was dissolved in 100 mL ethanol and cooled to 0–10 ◦C. Diazonium salt solution was added to the above solution at a temperature below 10 ◦C over a 15–20 min period, with the pH level maintained at 8–9 using a sodium hydroxyl solution. The reaction mixture was stirred for a further 3 h. Pyrazolone-1 was precipitated by adjusting the pH of the coupling solution to 4 using acetic acid, then collected and dried to get 3.5 g of crude product Pyrazolone-1. The crude Pyrazolone-1 was purified by flash chromatography on silica gel (eluting agent: ethyl acetate / petroleum ether = 1/100, v/v). Characterization: yellow solid; yield: 80.5%; <sup>λ</sup> (max) <sup>=</sup> 420 nm (in decane); <sup>ε</sup> <sup>=</sup>16119 L·mol−1·cm<sup>−</sup>1. IR (KBr, cm−1): 3058.90 (Ar–H); 2958.60 (–CH3); 2929.70 (–CH2 <sup>−</sup>); 2856.40 (–CH3); 1656.70 (C=O), 1560.40 (Ar); 1521.70 (Ar), 1456.10 (–N=N–); 1392.50 (–CH3); 1244.00 (C–N–C); 1091.60, 792.70. 1H NMR (500 MHz, CDCl3): 14.478 (s, 1H, O–H); 8.00–7.99 (d, 1H, Ar–NH–N=Ar); 7.81–7.76 (d, 2H, Ar–H); 7.61–7.59 (d, 1H, Ar–H); 7.52-7.49 (t, 1H, Ar–H); 7.46–7.43 (m, 2H, Ar–H); 3.65–3.63 (d, 2H, –NCH2 <sup>−</sup>); 2.66 (m, 1H,–CH–); 1.80 (m, 5H, –CH2 −); 1.24 (m, 12H, –CH2 <sup>−</sup>); 0.81–0.78 (m, 12H, –CH3). 13C NMR (500 MHz, CDCl3): 159.03, 149.97, 136.36, 134.07, 129.88, 128.75, 127.00-123.52, 121.34, 110.76, 47.84, 38.47, 32.95, 30.65, 28.76, 26.22, 25.45-22.28, 14.11, 10.74. MS (MALDI-TOF) (DIF) M/Z (%): 461.30 (M–H)<sup>−</sup>. Calc. for (C29H42N4O): 462.30.

Pyrazolone-2: A quantity of 1.97 g (197 g/mol, 0.01 mol) 4-aminodiazobenzene was dissolved in 30 mL of water and 3 mL (0.036 mol) of concentrated hydrochloric acid. After it had been rapidly cooled to 0–5 ◦C, a 5 mL solution containing 0.72 g (69 g/mol, 0.0105 mol) of sodium nitrite was added. Ehrlich reagent was used to detect the termination of diazotisation, and sulfamic acid was used to remove the residual nitrous acid. The coupling component 1-(2-ethylhexyl)-3-(2-ethylhexyl)-5-pyrazalone 3.08 g (308 g/mol, 0.01 mol) was dissolved in 100 mL ethanol and cooled to 0–10 ◦C. Diazonium salt solution was added to the above solution at a temperature below 10 ◦C over a 15–20 min period, with the pH level maintained at 8–9 using a sodium hydroxyl solution. The reaction mixture was stirred for a further 3 h. Pyrazolone-1 was precipitated by adjusting the pH of the coupling solution to 4.0 using acetic acid, then collected and dried to get 4.5 g crude product Pyrazolone-2. The crude Pyrazolone-2 was purified by flash chromatography on silica gel (eluting agent: petroleum ether and ethyl acetate). Characterization: yellow solid; yield: 87.8%; <sup>λ</sup> (max) = 416 nm (in decane); <sup>ε</sup> = 25335 L·mol−1·cm−1. IR (KBr, cm−1): 3103.20 (Ar–H); 2958.40 (–CH3); 2929.40 (–CH2 <sup>−</sup>); 2863.30 (–CH3); 1695.60 (C=O), 1617.00 (Ar); 1579.80 (Ar); 1521.90 (Ar); 1461.90 (–N=N–); 1379.20 (–CH3); 1226.20 (C–N–C); 1174.50 (C–N–C); 1120.82; 1044.32; 891.32. 1H NMR (500 MHz, CDCl3): 13.51 (s, 1H, O–H), 7.92–7.90 (d, 2H, *J* = 9.0 Hz, Ar–H), 7.83-7.82 (d, 2H, *J* = 7.5, Ar–H), 7.45–7.36 (m, 5H, Ar–H), 3.60–3.52 (m, 2H, –NCH2 −), 2.60–2.57 (t, 2H, *J* = 15.5, Ar–CH2 <sup>−</sup>), 1.79 (m, 1H, –CH–), 1.67 (m, 2H, –CH2 <sup>−</sup>), 1.23 (m, 10H, –CH2 −), 0.92-0.88 (t, 3H, *J* = 15.0, –CH3), 0.86-0.80 (m, 6H, –CH3). 13C NMR (500 MHz, CDCl3): 158.65, 152.72, 149.92, 143.49, 130.93, 129.13, 124.73, 122.82, 115.80, 47.67, 38.45, 32.94, 30.62, 28.77, 26.21, 23.41, 14.12, 10.72. MS (MALDI-TOF) (DIF) M/Z (%): 516.30 (M–H)−. Calc. for (C28H40N4O): 516.40.

#### **4. Results and Discussion**

The 1H NMR spectra for the target dyes recorded in CDCl3 are shown in Figure 2. The structures of dyes were confirmed by the presence of four distinct downfield signals characteristic of the aromatic rings. For Anthra-1–2, the structure of the anthraquinone core was confirmed by the presence of four distinct downfield signals ("H1", "H2", "H3", "H4"). The "H1", "H2", "H3", and "H4" protons of Anthra-1–2 exhibited one singlet, two multiplets and one doublet between 10.71 ppm and 6.93 ppm. For the substituents in the N-position, these two dyes displayed one common multiple peak "H5" between 3.22 and 3.14 ppm, corresponding to the methene protons adjacent to the imide group. The "H6" protons appeared as a singlet between 0.88 and 0.83 ppm. For the other pyrazolone dyes Pyrazolone-1–2, the "H1–4" and "H7–9" protons of Pyrazolone-1 exhibited three separate doublets, one triplet, and one multiplet between 8.00 ppm and 7.43 ppm. The "H1", "H2", "H3" and "H4" protons of Pyrazolone-2 exhibited three separate doublets and one multiplet between 7.92 ppm and 7.36 ppm. Two dyes: Pyrazolone-1–2, displayed one common multiple peak "H5" between 3.60 and 3.52 ppm corresponding to the methene protons adjacent to the imide group. Moreover, the common multiple

peak "H6" of those two dyes represented the protons on the methyl group, and the "H6" protons existed between 0.78 and 0.92 equivalent to the twelve protons which were obtained. The chemical values confirmed the structural correctness of the four dyes.

**Figure 2.** 1H NMR spectra of four synthesized dyes.

#### *4.1. Absorption Properties of Four Dyes*

As can be seen in Figure 3, the uv-visible absorption spectra and data of all the dyes were recorded in decane and were found in the range of 416–650 nm, showing color variations from yellow to cyan. For the anthraquinone dyes Anthra-1–2, the maximum wavelength number (λmax) value depended on the position of the alkyl-amino substituented on the anthraquinone ring. When two alkyl-amino groups were substituted on 1,4-sites, the λmax was 650 nm, while the λmax was 545 nm when they were substituted on 1,8-sites. For azo pyrazolone dyes Pyrazolone-1~2, the λmax value depends on the conjugate rings. The λmax of Pyrazolone-1, which is a mono-azo dye with a naphene group, is 416 nm. While Pyrazolone-2 is a bisazo dye with two benzene groups, the size of conjugate rings is a bit larger than Pyrazolone-1, thus the λmax values (420 nm) are a bit longer than for Pyrazolone-1.

**Figure 3.** UV-visible absorption spectra of four dyes.

#### *4.2. Photoelectric Response Properties of Dyes*

The EFD cell was fabricated using normal processes. The switching behavior of the EFD cells was studied under a certain dc voltage. The transmission spectra of EFD displays were tested under different voltages. The results can be seen in Figure 4. For Pyrazolone-1 as an example, we can clearly see the transmission changes as the voltage rose to 20 V, showing that the oil film starts to break up at this voltage. By increasing the voltage from 0 V to 40 V, the transmission value of EFD cell at maximum wavelength (420 nm) increases from 20% to 70%, showing that the cell could switch from bright yellow color (under 0 V) to the most transparent or white color (under 40 V). The EFD display could not be changed to an obvious one transparent to the naked eye until the voltage reached 32 V. And the transmission curves did not change greatly under the higher voltage because of the saturation of the oil contraction. The micron photos of EFD displays at 0 V (the OFF state) and 32 V (the ON state) are shown in Figure 4. The aperture ratio was achieved by calculating the occupying area of oil and white area in Photoshop software, and the aperture ratio could reach as high as 68.5%–75.0%.

**Figure 4.** Transmission spectra of EFD display formulated by four formulated dyes under different voltages. (Insert: micron photos of pixels filled with different oils). (**a**) 5.5% Anthra-1, (**b**) 7.7% Anthra-2.

#### *4.3. Photo-Stability Research of Dyes*

The photo-stability test of EFD oils were evaluated following the standard method 60068-2-9. The aging test of the display was carried under accelerated sunlight conditions for 100 h at 45 ◦C, which can translate to several years of application lifetime under normal conditions [19].

The change of absorbance curves under different irradiation time is shown in Figure 5. The decrease of dye absorbance after a period of irradiation test depends on the degree of dye degradation. Generally, we can see that dye degradation of Anthra-2 and Pyrazolone-2 are higher than Anthra-1 and Pyrazolone-1. After 100h irradiation, only 1.2% of Anthra-1 dyes were decomposed. The good photo-stability of Anthra-1 was attributed to the substitution of –NH at 1,4-sites of anthrquinone structure, which could form two six-membered rings with two carbonyl groups. This special structure

could strengthen the electron delocalization and decrease the electron density of potentially destabilizing groups, such as C–N and C–O bonds. In contrast, 23.9% of Anthra-2 decomposed after only 40 h of irradiation. The structure of Anthra-2 is a 1,8- sites of anthrquinone, while only one six-membered ring could be formed at the 1-site. As a result, the other substitution of –NH at the 8-site could not be stabilized and was prone to decomposing under irradiation. The initial anthraquinone dyes underwent a hydroxylation of the alkyl carbon atom adjacent to the left hand nitrogen atom. This resulted in a photo-bleached molecule with the original alkly group of the left hand alkl amine now substituted with the group –CHO [20].

**Figure 5.** Photo-stability test results of four formulated dyes. (Irradiation condition: 0.55 W/m<sup>2</sup> (340 nm); Temperature 45 ◦C). (**a**) 5.5% Anthra-1, (**b**) 7.7% Anthra-2, (**c**) 7.0% Pyrazolone-1, (**d**) 5.1% Pyrazolone.

For azo pyrazolone dye Pyrazolone-1–2, it could be seen that after 20 h of irradiation, the absorption values of Pyrazolone-1 and Pyrazolone-2 increased by 3.2% and 10%, respectively. This abnormal phenomenon could be explained by the transformation of cis-tran isomerization of azo group caused by light. However, when the irradiation time was extended to 100 h, the absorption increase ratio R% of Pyrazolone-1 was maintained at around 3.2%, while the absorption of Pyrazolone-2 decreased to 83.7%, showing that 16.3% of Pyrazolone-2 was decomposed and few Pyrazolone-1 molecules were decomposed. This difference could also be explained by the difference of structures. Pyrazolone-2 is a bisazo dye, and the azo group adjacent to the pyrrazole structure could form a six-membered ring, while the other isolation azo group could be easily destroyed by strong photo irradiation. Pyrazolone-1 is a monoazo dye, meaning the azo group adjacent to pyrrazole structure could form a six-membered ring and was stabilized. Thus, it is much more stable than Pyrazolone-2.

The color coordinates (x, y) of colored EFD cell in CIE1931 color space and color deviation Δ*E* were monitored along with the accelerating irradiation test. The results could be seen in Figure 6 and Table 1. With the exposure of irradiation time, the color coordinates (x, y) of the colored EFD cell in the CIE1931 color space changed, showing changes of the color. Among the four colored EFD oils, the color coordinates (x, y) of Anthra-2 had the greatest change from (0.38, 0.20) to (0.45, 0.32), showing that the color changed from bright magenta to brown. The Δ*E* of Anthra-2 was found to be as high as 130.5 after 100 h of irradiation. The color coordinates (x, y) of Pyrazolone-1 changed from (0.462, 0.468) to (0.460, 0.468), showing that the color did not change anymore. The Δ*E* of Pyrazolone-1 was 2.0, showing a small color change which can hardly be distinguished by the naked eye. The color coordinates (x, y) of Anthra-1 and Pyrazolone-2 changed from (0.172, 0.215), (0.478, 0.464) to (0.183, 0.228), (0.479, 0.466), respectively. The Δ*E* of Anthra-1 and Pyrazolone-2 were 7.7 and 3.7, respectively, showing a slight color change. These performances are highly relevant to their structure stability. Structure of Anthra-2 is the least stable and its color changed a great deal, while structure of Pyrazolone-1 is the most stable and its color changed minimally.

**Figure 6.** Variation of color coordinates (x, y) of oils in CIE1931 color space and Δ*E*. (Irradiation condition: 0.55 W/m<sup>2</sup> (340 nm); Temperature 45 ◦C). (**a**) 5.5% Anthra-1, (**b**) 7.7% Dye-2, (**c**) 7.0% Pyrazolone-1, (**d**) 5.1% Pyrazolone-2.


**Table 1.** Color change (Δ*E*) of four formulated dyes after accelerated irradiation.

(Irradiation condition: 0.55 W/m<sup>2</sup> (340 nm); Irradiation time: 100 h; Temperature 45 ◦C).

#### *4.4. E*ff*ect of Irradiation on Backflow Property of EFD Devices*

The tendency of the oil to flow back into the pixel at a constant voltage is called "backflow". The reduction oil-level contraction affects pixel aperture for the backflow of oil, which has an impact on the ability of the switch and the power consumption of the display. Since a static reading is the mode of the operation for an EFD display, a low "backflow" property oil will lead to the decrease of

the refresh rate for the display. Thus, the power consumption can be reduced because of its scaling with the refresh rate.

In Figure 7, it can be seen that before irradiation, the backflow time of four oils were around 50–400 s under 30 V DC (Direct Current) voltage, which is rather long and acceptable. The faster refresh speed for such a low backflow property will lead to a static image in an EFD display. A long backflow time can be attributed to the non-polarity of dye structures. The discrimination of backflow time for four oils is due to the difference of the non-polarity of their structures. When EFD cells underwent irradiation for a certain period, the backflow time of four dyes deteriorated sharply. For Anthra-2, the cell cannot be open after only 20 h of irradiation. This result might cause a dye's decomposition. When Anthra-2 was decomposed under irradiation, small fragments were released into oil, resulting in the rise of oil conductivity, thus the backflow time was extremely short and even pixels could not be opened. For Anthra-1 and Pyrazolone-1, in spite of the least dye decomposition, the backflow time also decreased from ~400 s and ~300 s to ~15 s and ~100 s, respectively after only 20 h of irradiation, and cells could not opened after 60 h of irradiation. These results showed that the effect of irradiation on the backflow property of EFD devices is not only relevant to a dye's structure, but also relates to the other materials, such as pixel wall materials and fluoropolymer materials. Any decomposition of these materials will release small high polarity fragments into oil, causing a rise in oil conductivity, and a decrease of the backflow time of EFD cell. This effect will be studied in-depth by our group, so it will not be extensively discussed in this paper.

**Figure 7.** Backflow properties of EFD devices with dyes under different irradiation times. (Irradiation condition: 0.55 W/m<sup>2</sup> (340 nm); Temperature 45 ◦C). (**a**) 5.5% Anthra-1, (**b**) 7.7% Anthra-2, (**c**) 7.0% Pyrazolone-1, (**d**) 5.1% Pyrazolone-2.

#### **5. Conclusion**

In summary, we have synthesized four EFD oil materials based on anthraquinone and azo pyrazolone dyes. The absorption, electrical-optical response, and photo-stability of these dyes were researched in detail. The dyes have good electrical-optical response properties due to their low polarity of whole molecules. However, the photo-stability was highly dependent on the structures. High polarity and high electronic density of the chemical bond are the photo-sensitive groups which are potentially destroyed under photo irradiation. Six-membered rings formed by hydrogen bonds could play an important role as a "protective structure", and they can highly increase the photo-stability property of these dyes. Irradiation could sharply decrease the backflow time of existing EFD devices, which is not only caused by the decomposition of oil materials, but also may be caused by the other materials in the pixels.

**Author Contributions:** Conceptualization, Y.D.; methodology, Y.D.; software, Y.D.; validation, Y.D. and S.L.; formal analysis, S.L.; investigation, H.J.; resources, Y.D.; data curation, H.J.; writing—original draft preparation, Y.D. and S.L.; writing—review and editing, Y.D.; visualization, D.Y.; supervision, B.T.; funding acquisition, G.Z. 'All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by National Key R&D Program of China (No. 2016YFB0401501), National Natural Science Foundation of China (Grant No. U1601651), Program for Changjiang Scholars and Innovative Research Team in University (IRT13064), Program for Science and technology project of Guangdong Province (Nos. 2018A050501013), National Natural Science Foundation of Guangdong, China (2018A0303130059), Science and Technology Program of Guangzhou (No. 201904020007, 2019050001), Guangdong Provincial Key Laboratory of Optical Information Materials and Technology (Grant No. 2017B030301007), MOE International Laboratory for Optical Information Technologies and the 111 Project.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article*
