1. Introduction

Domain engineering [1], whose target is to produce desired ferroelectric domain structure with the precise parameters in the ferroelectric materials, is an attractive technique for nonlinear optics and electro-optic devices. Recently more particular attentions are paid to the periodically poled lithium niobate (PPLN). Using the excellent nonlinear and electro-optic properties of PPLN, the research activities in the optical parametric oscillators [2,3,4], the nonlinear optical wavelength conversion [5], and the electro-optic controlled optical element [6,7,8], have been greatly revived. The more complex domain structure used for nonlinear photonic crystals is also studied [9]. Domain inversion can be achieved by the external electric-field poling which is currently the most reliable and preferred method of domain engineering. Besides the poling method used, the choice of dopant for LiNbO₃ is also very critical and attractive to optimize the physical property for domain inversion. 5 mol.% MgO-doped LiNbO₃ (MgO:LiNbO₃) crystal has been investigated and developed because of its higher resistance to photorefractive damage than that of congruent LiNbO₃, and it has been proved by the experiments that MgO:LiNbO₃ has a larger nonlinear coefficient and a lower coercive field (4.45 kV/mm) than those of congruent LiNbO₃ [10]. The periodic poling of the MgO:LiNbO₃ has also been achieved [11,12]. The domain reversal by pulsed-field in congruent grown LiNbO₃ doped with ZnO (ZnO:LiNbO₃) at concentration larger than 5 mol.% has also been analysed, and the polarization switching field has been found to decrease significantly with the doping. To 8 mol.% ZnO doped LiNbO₃ the threshold field and internal field are as low as 2.5 and 0.5 kV/mm, respectively[13].

Electrochromism is the reversible change in optical property which can occur when the electrochromic material is electrochemically oxidized or reduced, and is firstly found in the transition-metal oxide films such as tungsten oxide film [14] and iridium oxide film [15]. Commonly, the electrochromic material can display distinct visible color change which is between a transparent (“bleached”) state and a colored state, or between two colored states. This definition can be extended to more extensive spectral range. Electrochromism has been applied for the ultrafast color display [16] and many other commercial applications. We have already in experiments found the electrochromism during the ferroelectric domain inversion in congruent z-cut LiNbO₃ crystals codoped with 0.15 wt.% Fe₂O₃ and 0.18 wt.% RuO₂ (Fe₂O₃:RuO₂:LiNbO₃) at room temperature [17]. The electrochromism occurs accompanying the domain inversion simultaneously in the exact same poled area. But the mechanism of this effect is not clear, and we even can not determine which dopant plays the dominant role in the effect.

Being a transition-metal oxide, RuO₂ is potentially applied for the electrochromism. It has been reported on the electrochromic behavior, which is induced by supercapacitor charging/discharging, in an amorphous hydrous RuO₂ thin films electrode [18]. In this paper, we report for the first time the electrochromism accompanying the ferroelectric domain inversion in congruent 0.25 wt.% RuO₂-doped z-cut LiNbO₃ (RuO₂:LiNbO₃) crystals, which is completely reversible when the domain is inverted from the reverse direction, at room temperature. The electrochromism is proved to accompany the domain inversion simultaneously and is localized to the same poled area as the domain inversion. The properties, both simultaneity and localization, are proved solidly by the real-time visualization of domain pattern with coherent light, the real-time measurement of optical transmittance change and poling current, and the micrographic analysis after being etched in hydrofluoric acid. The influences of electric field and annealing treatment are also studied in this paper. The results are compared with those of congruent LiNbO₃ and Fe₂O₃:RuO₂:LiNbO₃ crystal [17].

2. Experiment and result

2.1 Experiment set-up

In the following experiments, the z-cut, 0.5-mm-thick, double side polished RuO₂:LiNbO₃ substrates are diced into 15mm×15mm squares and have a approximate area of 36mm² (6mm×6mm) contacting the liquid electrodes (water) which provide the homogeneous electric field. The LiNbO₃ crystals used are grown by the Czochralski technique from the congruent melt, and Ruthenium is introduced into the melt in the form of RuO₂. The RuO₂ doping level is 0.25 wt.%, and the color of the virgin sample is pale-yellow. A high voltage supply is used for supporting pulse or continue external field with 20kV and 2mA maximum output. The switching current caused by the charge redistribution within the crystal is monitored by a micro-amperemeter. Imaging of the transmitting coherent optical field at the wavelength of 633nm through a lens is captured by a charge-coupled-device (CCD) camera and processed by a computer. The transmission spectrums of different states are measured by a spectrometer. During the poling cycle, the real-time measurement of optical transmissivity modulation is performed by detecting the optical intensity change of the collimated light illuminating over the entire electrode area at 514nm wavelength. The domain pattern after being etched is observed with a conventional microscope. The scheme of the experimental facility is shown in Fig. 1.

 

Fig. 1. Schematic view of the experimental facility

Download Full Size | PDF

2.2 Domain inversion by external electric-field poling

It is well known that the ferroelectric hysteresis loop of LiNbO₃ is asymmetric which indicates the existence of the internal field [19]. It induces the inequality of the coercive field along both directions of the crystals. In this paper, the direction along which the external field inverts the domain at higher field strength is named “forward direction”, and the opposite direction along which the external field inverts the domain at lower field strength is named “reverse direction”. The corresponding domain inversion process is named “forward poling” and “reverse poling”, respectively. The virgin sample is first poled along the forward direction, and then along the reverse direction. A complete poling cycle is composed of two successive domain inversions including a forward poling and a reverse poling.

The influence of the electric field on the domain inversion is investigated. During the poling process the collision ionizations, which can generate the electron avalanche if the external electric-field is large enough, occur in the crystal structure. The energy absorbed by the electrons can be transformed to thermal energy by electron-phonon collision, which can breakdown the crystal wafer by generating abrupt temperature increment. Commonly, in order to avoid gaining sufficient energy to destroy the sample, the pulsed field and thin wafer are adopted [20]. In our experiments we find that, to 0.5mm-thick RuO₂:LiNbO₃, the pulsed electric-field is not necessary for domain inversion, and can be replaced by the continuous one if the applied voltage ramps up slowly enough (<50V/s). This property is different from that of congruent lithium niobate. We ascribe it to the suppression of collision ionization induced by the occupation of Ru ion within the crystal structure which inhibits the electron avalanche breakdown eventually. In the following experiment, the continuous electric-field is used to invert the domain. The external voltage starts from 0V and ramps up linearly at a rate of 50V per second.

When the external electric-field reaches or exceeds the coercive field, the domain inversion starts. In order to measure the coercive fields in both directions, we monitor the displacement current and the imaging of the transmitting coherent optical field in the poling progress. The moment when both current and domain pattern appear is determined, and the electric-field applied at this time is defined as coercive field. The measurements are repeated ten times respectively and the both averages are used. The experiment results reveal that in the forward direction an applied field needs to be approximately 19.6kV/mm to initiate the domain inversion, and in the reverse direction the value is 15.2kV/mm. So the internal field is 4.4kV/mm.Compared with congruent LiNbO₃, the coercive fields and internal field of RuO₂:LiNbO₃ are not found to decrease substantially [21]. We think that while Ru ion can occupy the vacancy in the lattice, but this substitution in the lattice site can not suppress the nonstoichiometric defect which is believed to be responsible for the high value of coercive field. So the doping of the Ru ion can not decrease the coercive fields of RuO₂:LiNbO₃.

2.3 Electrochromism accompanying domain inversion

During the domain inversion process, the significant color changes of the crystal are observed by naked eyes: the poled area (central area) is “bleached” during the forward poling (shown in Fig. 2), and is “colored” again in the following reverse poling. The colour of crystal recovers after a poling cycle and there is no contrast between central area and edge area.

After domain inversion, the transmission spectrum from 300nm to 700nm wavelength is measured by the spectrometer which further proves the presence of electrochromism in the poled area. Shown in the Fig. 3 is the comparison of the transmission spectrum among three different states during a poling cycle: virgin state, after forward poling and after reverse poling. After forward poling, there is a shift to the short-wave band of the transmission spectrum from 460nm to 600nm wavelength, and a shift to the long-wave band from 300nm to 460nm wavelength. There is an intersection point at approximately 460nm wavelength. At about 530nm wavelength, there is a maximum relative quality between virgin state and after forward poling state, and the modulation of transmittance is about 10%. The experiment result also proves that there is no more change of the transmission spectrum from 700nm to1000nm wavelength. Obviously the electrochromic behavior is opposite in the both spectral range, which is different from the one of Fe₂O₃:RuO₂:LiNbO₃ [17]. The transmission spectrum of the virgin crystal is almost superposed with the one after a complete poling cycle, and it indicates clearly that the change process is completely reversible after a complete poling cycle.

 

Fig. 2. Observation of the crystal after forward poling. The poled area (central area) is “bleached” during the forward poling, and correspondent with the shape of liquid electrode which is distorted a little by extrusion of the silicon gaskets. After reverse poling the poled area is “colored” again and there is no contrast between central area and edge area

Download Full Size | PDF

 

Fig. 3. Comparison of the transmission spectrum among three different states during a poling cycle: virgin state, after forward poling and after reverse poling.

Download Full Size | PDF

In order to study the influence of annealing conditions, the substrate is annealed at 150°C for some time after the forward poling. With the increase of annealing time, the “bleaching” phenomenon in the poled area disappears and no optical modulation is observed. There is an integral shift to the long wavelength of the transmission spectrum with the increase of annealing time. The change of transmission spectrum induced by annealing treatment is shown in Fig. 4.

 

Fig. 4. Change of transmission spectrum induced by annealing treatment at 150 °C for different time. The comparison is among virgin state, after poling, after poling and annealing for 18 hours, and after poling and annealing for 32 hours

Download Full Size | PDF

During the poling process, the real-time imaging of transmitting coherent optical field at 633nm wavelength through a lens, which directly indicates the move of domain wall [22], is visualized. The 633nm wavelength is selected while both 633nm and 514nm wavelength can be used for the measurement actually. The CCD camera continuously records the evolving domain pattern with the temporal revolution 20 frames/s. Shown in the Fig. 5 is the one of the image (part) of the domain wall with coherent light. The boundary of the domain has the hexagonal shape (showed by the dashed line) and is expanding outside. The mechanism can be explained by the linear electro-optic, piezoelectric and interference effect [23]. The refractive-index modulation induced by linear electro-optic is presented as Δn ~ γ₁₃E₃ where γ₁₃ is the electro-optic tensor whose sign depends on the orientation of the c axis and E₃ is the external along the c axis. So there is a refractive-index change equal to 2Δn at the boundary of the two areas with opposite polarity. The piezoelectric thickness change 2Δd is generated by the applied voltage. Taking both linear electro-optic and piezoelectric effects into consideration, when an incident plane wave propagates along the domain boundary the phase shift ΔΦ, which is presented as ΔΦ=2π/{2Δnd+2(n₀-n_w)Δd}, is induced between two waves at both sides of the boundary and generates interference patterns eventually. The phase shift can be altered by changing the electric field, so the domain boundary appears and disappears at some interval of the external field. The visible variation of the contrast at the domain boundary is observed in the experiment. Because the lens is moved out of focus, the pattern of the domain boundary is broader, but the revolution is lower than that at the focus. It is proved by the real-time visualization of the domain wall and the observation of the color change in a special region that the electrochromism coincides with the domain inversion, and is localized to the exact same area. But the evidence is not enough yet. The important properties, localization and simultaneity, can also be proved by the following two experiments.

 

Fig. 5. Real-time visualization of the transmitting coherent optical field at 633nm wavelength through a lens. The hexagonal shape discontinuity (marked by the dash line) represents the domain boundary. The domain boundary is expanding outside

Download Full Size | PDF

We interrupt the poling process while the image of domain wall is still visible, and the domain pattern will be “frozen” within the crystal. After being etched in the hydrofluoric acid for 24 hours, the domain pattern is transformed to the surface topography which can be visualized by optical microscope in reflection mode. On the z+ surface the distinct domain boundary between two areas with opposite polarization state and different color is shown in Fig. 6. The left side is the domain inversion area which is “bleached” and depressed by etching, and the right side is the un-inverted area. The finer structure of the domain wall is depicted in Fig. 7, in which the hexagonal shape of domain boundary is obvious, and the dimension magnitude of the domain is 1μm. The results prove solidly that the electrochromism is exactly localized to the domain inversion area.

 

Fig. 6. Domain boundary between two areas with opposite polarization state and different color after being etched in hydrofluoric acid. The left area is poled and bleached, and the right area is not poled

Download Full Size | PDF

 

Fig. 7. Finer Structure of ferroelectric domain with hexagonal shape boundary after being etched in hydrofluoric acid. The dimension magnitude of domain is 1 μm.

Download Full Size | PDF

During the poling cycle, the real-time measurement of color change is performed by detecting the transmission power change of collimated light at 514nm wavelength which is illuminating the entire electrode area. The 514nm wavelength is selected because it is close to the 530nm and there is a distinct change of transmittance at this wavelength shown in the transmission spectrum. At the meantime, the switching current caused by the charge redistribution within the crystal, which indicates the occurrence of the domain inversion, is monitored by a micro-amperemeter. The forward poling is performed with the constant voltage of 9.8kV, and the reverse one is performed with the constant voltage of 7.6kV. The results are shown in Fig. 8(a) and Fig. 8(b), respectively. The results show that the congruent 0.25 wt.% RuO₂:LiNbO₃ exhibits a 4%-5% modulation of optical transmittance at 514nm wavelength accompanying the domain inversion.

 

Fig. 8 (a). Transmissivity change of collimated light at 514nm wavelength which is illuminating entire electrode area (square dot) and the change of poling current (circle dot) during forward poling with constant voltage of 9.8kV

Download Full Size | PDF

 

Fig. 8 (b). Transmissivity change of collimated light at 514nm wavelength (square dot), and change of poling current (circle dot) during reverse poling with constant voltage of 7.6kV

Download Full Size | PDF

When the poling current appears, the intensity of the transmitted light varies simultaneously. When the poling current disappeared, the variety of the transmitted light intensity ended at the same time. The results above prove solidly that the electrochromism coincides with the ferroelectric domain inversion. The current spikes are corresponding to the discrete domain wall motion with the velocity varying from nearly zero to a maximum value, which is also observed from the real-time visualization of the domain wall. The process of poling current change is different from the one of Fe₂O₃:RuO₂:LiNbO₃ and the difference is maybe ascribable to the less density of pinning centers in RuO₂:LiNbO₃ compared with Fe₂O₃:RuO₂:LiNbO₃. We can also find that the large current pulse is corresponding to the maximum rate of color change. The increasing current amplitude is the evidence of the domain wall motion acceleration, and the domain wall motion is also in correspondence with the change of areas with different colors.

The experimental results prove solidly the relations, both simultaneity and localization, between the electrochromism and the domain inversion, and also prove that no electrochromic effect occurs if there is no domain inversion. The results are consistent with those of Ref. 17.

2.4 Analysis and assumption

The mechanism of the special relations between the electrochromism and the domain inversion has not been clear, and we propose some analyses and assumptions. The ruthenium exiting in the virgin LiNbO₃ crystal possesses three different valence states [24]: Ru³⁺, Ru⁴⁺, and Ru⁵⁺, in which the Ru⁴⁺ is the dominant state. Ru⁴⁺ can be reduced to Ru³⁺ by incorporating an electron or be oxidized to Ru⁵⁺ by losing an electron.

When the domain is inverted in the forward poling, a displacement current flows in the external circuit because the charges are redistributed, and the action of external field makes electrons move directionally within the crystal structure. In this process Ru⁴⁺ is reduced to Ru³⁺ by the following transition simultaneously:

Here e^- denoting an electron. So the Ru³⁺ sites are generated to producing the “bleaching” effect upon the electrochemical reduction during the forward poling.

When the domain is inverted in the reverse poling, the electrons move directionally within the crystal structure again in the reverse direction and Ru³⁺ is oxidized to Ru⁴⁺ by the following transition at the same time:

Upon the electrochemical oxidation and reduction the Ru⁴⁺ sites are generated to producing the “coloring” effect. The intervalence electron transference between Ru⁴⁺ and Ru³⁺ plays a key role in the spectrum shift within different spectral range by the change of the photon absorption.

The above process creates the electrochromic behavior which is completely reversible during the poling cycle. We think the entire poling device including crystal and electrodes to be a “capacitor”, and think the poling cycle including forward poling and reverse poling to be a “charge and discharge” cycle. During the domain inversion, the total delivered charge is Q_DI. It is presented as Q_DI = 2P_sA where P_s is the spontaneous polarization of the crystal and A is the poled area. Q_DI is a constant in our experiments, and has the magnitude 10^-5C. To the accompanying electrochromism, the total delivered charge Q_E is determined by Q_DI. So the electrochromism is induced and controlled by domain inversion. Because the total delivered charge is holding constant during the charge and discharge cycle, the induced electrochromic effects are reversible completely as shown in Fig. 3. During annealing treatment the valence electron is stimulated, and more and more Ru ion is oxidized to Ru⁵⁺. So the optical transmittance spectrum shifts to the long-wave band as shown in the Fig. 4.

So we think that the electrochromism is induced by domain inversion, and accompanies domain inversion simultaneously in the same poled area during the electric poling process. The charge redistribution within the crystal structure caused by domain inversion is the source for electrochemically oxidation and reduction of Ru ion to produce the electrochromic effect.

3. Conclusion

In conclusion, we report the electrochromism phenomena accompanying the ferroelectric domain inversion in congruent 0.25 wt.% RuO₂-doped z-cut LiNbO₃ crystals at room temperature. During the electric poling process, the electrochromism accompanies the ferroelectric domain inversion simultaneously, and is localized to the same poled area as the domain inversion. The RuO₂:LiNbO₃ exhibits a 4%-5% modulation of optical transmittance at 514nm wavelength accompanying the domain inversion. The electrochromism is completely reversible when the domain is inverted from the opposite direction. The relations, both simultaneity and localization, between electrochromism and domain inversion are proved solidly by the experiments. The coercive fields and internal field of RuO₂:LiNbO₃ are not found to change substantially. The influences of the electric field and annealing conditions on domain inversion and electrochromism are also discussed. We provide the reasonable assumptions and analyses that the charge redistribution within the crystal caused by domain inversion is the source for electrochemically oxidation and reduction of Ru ion, and the intervalence electron transference between Ru⁴⁺ and Ru³⁺ plays a key role in the spectrum shift within different spectral range by the change of the photon absorption. The real mechanism is more complicated and needs further research.