We present video footage demonstrating real-time visualization of domain formation in periodically-poled lithium niobate (PPLN). This in-situ, non-destructive technique provides important visual information concerning the global quality and dynamics of domain patterning during the fabrication of PPLN.
©2000 Optical Society of America
In the past five years, patterned domain reversal in lithium niobate has been extensively investigated using electric-field poling techniques [1–7]. The impressive nonlinear properties and well-established fabrication procedures have made periodically poled lithium niobate devices a staple of the literature in recent years [8–14]. Although a variety of methods have recently been used to improve the pattern fidelity of the reversed domains [4–7], the basic technique of domain patterning in PPLN remains unchanged. To our knowledge, the widespread fabrication of PPLN still uses a lithographically patterned insulating layer, lithium chloride liquid electrodes, and a chuck sandwiching the sample between two o-rings — all features used in the early demonstrations of patterned poling.
2. Control of Domain Spreading
Since the advent of the electric-field poling technique, lateral domain spreading has continued to be perhaps the single most important aspect of domain patterning in periodically-poled lithium niobate. Upon application of an electric-field exceeding the coercive field of the material, domains first nucleate in regions underneath the electrodes, and subsequently spread laterally underneath the adjacent insulating layer [1–3]. Under most poling configurations, if the electric-field is not removed after an appropriate duration, the reversed domains continue to propagate under the insulating layer, and eventually merge together forming one completely reversed domain with no patterning [1,3]. The initial method of determining the poling progress — and the method most widely used to date — simply measures the current flowing across the sample [2,3]. By knowing a priori the total area to be poled, A, and the spontaneous polarization, Ps=78 µC/cm2 for LiNbO3, the total charge to be deposited on the sample, Q=2APs, is easily calculated . In order to deposit the proper amount of charge, a poling current I must be applied for the proper duration T such that IT=Q=2APs. In general, however, the specific poling current, I, depends on the applied voltage, the coercive field, and the sample thickness, so that even if the same voltage is always used, the poling current can change slightly from sample to sample. As a result, poling the sample with a single voltage pulse for a specified time almost always provides a total poling charge different than the desired charge Q. As a result, researchers often apply several voltage pulses, using the measured poling current from previous pulses to determine the duration of the last few pulses. This method has proved satisfactory over the years.
Alternatively, if the poling configuration has a current-limiting series resistor, the poling current can be limited to levels such that the total poling process takes on the order of several seconds or even minutes. This “slow poling” technique has the advantage that the poling current can be acquired and integrated in real-time using an A/D card and appropriate software on a computer. If the same A/D card is used to send the poling voltage pulse to the high-voltage amplifier, the software that monitors the integrated current can also switch the poling voltage off once the desired charge has been deposited on the sample. Because of this, the entire poling charge can be applied in a single pulse. This is the method of fabrication that we have used to date.
Another method of achieving the proper charge deposition was developed by Miller et al. in their effort to demonstrate good domain fidelity in 6 µm period PPLN . The authors replaced the insulating layer of photoresist with a layer of spin-on-glass and thoroughly analyzed the specific details of the electric-field shielding under the insulating layer. They then demonstrated that a fortuitous choice of patterned duty cycle and applied voltage could limit the lateral domain growth under the spin-on-glass. Since the depth of spreading under the insulating layer was not controlled by the duration of the electric field as before, the poling voltage was simply left on until no further poling current was measured.
Batchko et al. Reported backswitch poling; a modification of Miller’s technique utilizing ferroelectric spontaneous backswitching [6,17]. In this approach, the domains are grown using NiCr electrodes and a photoresist insulator and are allowed to spread until reaching a duty cycle of approximately 100%. Once the domains begin to merge together beneath the photoresist, the applied voltage is rapidly lowered, thus allowing the onset of backswitching. During backswitching, the domain walls move in the reverse direction, i.e., back toward the electrodes. Poling is terminated by again raising the voltage. As with the previous approach, the deposited charge can integrated and backswitching is then terminated once the domain duty cycle reaches a specified value. Due to the high nucleation density of the backswitching process, high-quality short-period domain structures can be patterned using this process. Using backswitch poling, they fabricated full 76-mm-diameter 0.5-mm-thick 4-µm-period PPLN wafers and demonstrated SHG of 460-nm light.
3. Internal Strain
Given the above survey of poling techniques, it is clear that achieving the desired patterning requires accurate knowledge of both the spontaneous polarization and the area to be poled. While it seems that the area to be poled should be very simple to calculate — for a single grating the area only requires knowledge of the o-ring radius — in fact quite often such calculations lead to slightly underpoled crystals. Using the current-limited poling setup, we observed that the samples often left small regions of unpatterned material. One in-situ method of ensuring that such regions are eliminated is to watch for a drop in the poling current and a corresponding rise in the poling voltage — both effects indicate that the entire sample has completely poled under the electrodes and is now spreading under the insulating layer . Another method is to monitor second-harmonic generation through the crystal during poling . However, aside from these methods, there have previously been no in-situ technique to offer a full visualization of the domain formation in the crystals.
Very early in the development of the electric-field poling technique, researchers noticed that after poling patterned LiNbO3, the sample exhibited significant scattering when viewed through crossed polarizers. The strength of the scattering effect is dependent on the orientation of the crystal with respect to the crossed polarizers, and can be annealed away by heating the crystal at 200 °C for several hours. Additionally, the crystal self-anneals at room temperature after a period of several days. The effect is most likely due to birefringence associated with oppositely-oriented adjacent domains, however the fields that cause the birefringence have not been elucidated. Since the scattering effect only occurs at locations of oppositely oriented domains, it can be used as a visual indicator of the poling quality of a sample. For example, a single domain sample exhibits no scattering, a partially patterned sample exhibits regions of scattering where domain reversal has taken place, and a completely patterned (i.e. with a 50% duty cycle) sample exhibits scattering everywhere inside the region defined by the o-rings of the poling fixture. Qualitative information about the poling quality can be obtained by viewing the sample though crossed polarizers under a microscope. Such a diagnostic is useful since the true measure of the poling quality can be determined only by removing the photoresist — thereby prohibiting further poling the sample - and etching the sample in HF acid to reveal the true domain structure. Figure 1 shows a partially poled sample of a multi-grating PPLN sample before and after etching in HF acid. The sample was fabricated using photoresist trenches to define the desired periodic poling pattern, and was poled using standard techniques.[1,2] The top-left image was recorded by viewing the sample (1.25x magnification) through crossed polarizers. This image was taken after the poling process, but before removing the photoresist and etching the surfaces in HF acid. The top-right image was recorded using the same magnification, but without the crossed polarizers, after removing the photoresist and etching the sample in HF acid. Furthermore, the top-left image (pre-etch) was recorded with the illuminating light passing through the crystal, so that the internal strain could be seen. In contrast, the top-right image (post-etch) was recorded with the illuminating light reflecting off of the top surface of the crystal so that only the etched surface could be seen. Fig. 1 shows that all of the domain structure information contained in the post-etch image (top-right) is also contained in the pre-etch image (top-left) as well. Furthermore, as noted above, the pre-etched sample can be subjected to further processing, whereas the post-etched sample cannot.
Although viewing the sample microscopically through crossed polarizers provides significant information concerning the poling quality, the sample can not be microscopically viewed in situ. If the sample is to be further poled, it needs to be inconveniently reloaded into the poling fixture. Furthermore, at this point, the charge calculation becomes extremely difficult to accurately predict for two reasons. First, it is very difficult to calculate the total unpatterned area scattered throughout the crystal. Second, even supposing the area could be accurately calculated, such a calculation is not necessarily useful since the original charge calculation evidently proved insufficient to completely pattern the sample in the first place.
We note here that the stress induced scattering exhibited by PPLN is not exhibited by other materials, such as BaTiO3 and RbTiAsO4, and is significantly reduced in LiTaO3. In addition, the domain formation is most clearly observed when using photoresist trenches. However, the effect can still be observed when the photolithographic pattern is formed using photoresist over metal lines, as is the case in current short-period poling procedures. The moderate duty cycle of the metal lines allows sufficient light to pass through the crystal to observe the effect described here, although the contrast is usually weaker and the effect is generally harder to see.
4. Real-time in-situ monitoring
In light of this, we attempted to visualize the domain formation in periodically-poled lithium niobate by placing the sample and Plexiglas poling fixture between crossed polarizers during the poling process. In this way the progress of the domain formation in the crystal could be monitored in-situ in real time. With a low-current configuration, this method can be used to provide a visual cue to manually terminate the poling voltage pulse. In a high-current poling configuration, this method loses its real-time monitoring advantage, but still provides valuable in-situ post-pulse visual information of the domain formation. Such post-pulse data can be used to determine whether or not the entire crystal has been poled, so that, if necessary, additional pulses can be applied to the crystal before removing from the poling fixture.
The sample was illuminated by a fluorescent light source and imaged by a CCD camera. The video was recorded on a VCR, and was later converted to QuickTime 4.0 movies using a frame grabber and a software package. We show five movies below, demonstrating the power and utility of this visualization technique.
The first video (Fig. 4) shows the domain formation in a full-wafer (3-in. diameter) 28.5-µm period single-grating sample. The footage shown in this video is perhaps the best we have recorded. There are several points of interest. First, the poling starts at the edge of the sample near the o-ring and propagates inward, indicating the presence of large fringing fields and high piezeoelectric stress at the o-ring edges. As the poling progresses, the domain formation is clearly visible from the domain scattering that sweeps through the crystal. At the end of the movie — where the voltage has already been turned off — there are still some small unpatterned regions in the center of the crystal. Using this visualization technique, these unpatterned regions are very obvious, and a quick application of a short second pulse was sufficient to pole these last regions. In fact, we simply used the visual cues to determine when to manually terminate the second pulse, rather than rely on a charge calculation. Because of the current-limited poling technique used during this process, the total time elapsed during the video was approximately 140 seconds. In order to compress the video data and to better visualize the domain formation, the QuickTime movie is played at approximately thirty-five-times the real speed. If desired, the real-time poling duration could be shortened simply by reducing the current-limiting series resistor to increase the poling current.
As this video shows, domain nucleation does not occur uniformly across the device. Furthermore, whole sections of the crystal are poled before other sections have even been nucleated. However, the domains which pole first eventually begin to spread slightly under the insulating photoresist pattern. At this point, the local coercive field required to push this domain further under the photoresist pattern exceeds the coercive field in the unpoled regions of the crystal. At this point, the lateral spreading of the formed domains stops until the rest of the crystal has been poled. In this way, the areas in which domain reversal first occurs do not have a significantly different duty cycle than the areas in which domain reversal occurs last.
The second video (Fig. 5) shows the domain formation in a 14-mm diameter circular region of a multi-grating crystal (Λ=16-20 µm). For reasons not discussed here, rather than pole the entire crystal at once, we poled small portions of the sample using 14-mm diameter o-rings. The first frame of the video footage shows that the area currently being poled slightly overlaps a region on the right that has already been poled. Careful investigation of the video shows that the poling primarily starts both at the o-ring edge and at the edges of the multi-grating segments. Both are locations where fringing electric fields are strongest and cause higher nucleation rates than in the rest of the crystal. Again, the video clearly shows the domains forming as the crystal is poled. Careful observation of the last few frames also reveals that the crystal relaxes as the high-voltage pulse is terminated. This is seen as a slight dimming of the scattered light. When the poling is complete, the difference between poled and unpoled regions is very prominent. The unreversed single-domain separation zones between individual gratings and the larger separation zone between different multi-grating crystals (near the top of the o-ring) are clearly visible due to the lack of scattering in this region. The real-time poling of this sample lasted approximately 5.75 seconds, so that the QuickTime movie plays at almost exactly the real-time speed.
The following two videos (Fig. 6) show the domain formation in two more samples. The first sample consisted of a 28.5-µm period single-grating full wafer with triangular and trapezoidal sections shielded from the electric field by photoresist, so that they remained unreversed during the voltage pulse. The domain formation of this wafer again demonstrates the progression of poling from the edges of the sample to the center and reveals good discrimination between periodically poled sections and unpatterned sections of the crystal. The video also reveals interesting internal patterns on the left- and right-hand sides of the crystal. The fourth video shows the domain formation in three PPLN fan gratings (Λ=25.5-31.25 µm) . This video shows strong initial nucleation of several areas along a diagonal opposite to the direction of the fan grating. Again, interesting internal strain structures are observed. These internal strain effects do not appear to have any significant effects on the nonliner properties of the crystals. The total duration of the poling voltage pulse for both wafers was approximately 140 seconds, and again the QuickTime videos are both played at approximately thirty-five times real speed.
The next video (Fig. 7) shows domain formation in a full wafer of multi-grating crystals, again revealing domain nucleation at the tips of the individual gratings. This video was one of the first that we acquired, and used a small flashlight as an illuminating source, so that only a portion of the crystal was well illuminated.
The final video shows a multi-grating sample (Λ=26.0-31.2 µm) being poled beyond the desired 50% duty cycle. The sample was intentionally overpoled to illustrate that the visible internal strain disappears as the sample returns to being single domain. The image acquisition and sequence of figures are identical to those of Fig. 1. The video again shows the domain formation sweeping across the crystal, as evidenced through the scattering of the inspection light. As the poling progressed past the ideal 50% duty cycle, the intensity of the scattered light uniformly decreased. The scattering phenomena altogether ceased in areas where domains began merging together to form large single-domain reversed regions. The video also shows that the portions of the crystal which poled first also overpoled and became entirely single domain first. Figure 9 shows images of the pre-etched and post-etched sample of Fig. 8. Similar to Fig. 1, the pre-etch image of Fig. 9 reveals the location of domain walls with areas of bright scattering. The scattering is less bright than that of the corresponding image in Fig. 1, indicating a large asymmetry in the duty cycle. The expanded etched image confirms that the dark areas in the pre-etched image correspond to large single domain regions, while bright areas correspond to areas with anti-parallel domains. In addition, note that in the areas where periodic domains do still exist, the duty cycle is almost unity, indicating that the overpoling occurs fairly uniformly over the whole crystal.
We finally point out that observing the domain formation in PPLN is very easy to do as a human observer, but is considerably harder to actually capture on video due to contrast, lighting, vantage point, and dynamic range issues. The contrast and visibility variation in the videos presented here should be noted. Even in the best footage (Fig. 4.), the domain formation is hard to observe in the four corners of the crystal, due to the uniaxial nature of the crystal, although all portions are easily visualized (though not simultaneously) by changing the vantage point.
In conclusion, we have demonstrated an effective technique for visualizing the domain formation in periodically poled lithium niobate. The charges trapped at the domain boundaries cause scattering of polarized light, and can therefore be observed by viewing the sample through crossed polarizers. By using a low poling current, we were able to record the domain formation in real time using a only a CCD camera and a VCR. This in-situ, non-invasive technique allows the user to investigate the poling dynamics, and to determine the global quality of the domain formation in the crystal. This technique can be used along with the recorded current and voltage traces to determine whether or not a sufficient amount of charge has been deposited on the sample. With a low-current configuration, this method can be used to provide a visual cue to manually terminate the poling voltage pulse. In a high-current poling configuration, this method loses its real-time monitoring advantage, but can still provides valuable in-situ post-pulse qualitative data. This technique offers important information regarding the quality of domain formation in the crystal, and could serve as a powerful in-situ quality control mechanism if periodically poled lithium niobate makes the technological leap into the forum of mass production.
References and links
1. M. Yamada, N. Nada, M. Saitoh, and K. Watanabe, “First-order quasi-phase matched LiNbO3 waveguide periodically poled by applying an external field for efficient blue second-harmonic generation,” Appl. Phys. Lett. 62, 435 (1993). [CrossRef]
2. L. E. Myers, R. C. Eckardt, M. M. Fejer, R. L. Byer, W. R. Bosenberg, and J. W. Pierce, “Quasi-phase-matched optical parametric oscillators in bulk periodically poled LiNbO3,” J. Opt. Soc. Am. B 12, 2102 (1995). [CrossRef]
3. V. Pruneri, J. Webjoorn, P. St. J. Russell, J. R. M. Barr, and D. C. Hanna, “Intracavity second harmonic generation of 0.532 µm in bulk periodically poled lithium niobate,” Opt. Commun. 116, 159 (1995). [CrossRef]
4. M. Reich, F. Korte, C. Gallnich, H. Welling, and A. Tünnermann, “Electrode geometries for periodic polling of ferroelectric materials,” Opt. Lett. 23, 1817 (1998). [CrossRef]
5. G. D. Miller, R. G. Batchko, W. M. Tulloch, D. R. Weise, M. M. Fejer, and R. L. Byer, “42%-efficient single-pass cw second-harmonic generation in periodically poled lithium niobate,” Opt. Lett. 22, 1834 (1997). [CrossRef]
6. R. G. Batchko, V. Y. Shur, M. M. Fejer, and R. L. Byer, “Backswitch poling in lithium niobate for high-fidelity domain patterning and efficient blue light generation,” Appl. Phys. Lett. 75, 1673 (1999). [CrossRef]
7. P. T. Brown, G. W. Ross, R. W. Eason, and A. R. Pogosyan, “Control of domain structures in lithium tantalate using interferometric optical patterning,” Opt. Commun. 163, 310 (1990). [CrossRef]
8. W. R. Bosenberg, A. Drobshoff, J. I. Alexander, L. E. Myers, and R. L. Byer, “93% pump depletion, 3.5-W continuous-wave, singly resonant optical parametric oscillator,” Opt. Lett. 21, 1336 (1996). [CrossRef] [PubMed]
9. G. W. Ross, M. Pollnau, P. G. R. Smith, W. A. Clarkson, P. E. Britton, and D. C. Hanna, “Generation of high-power blue light in periodically poled LiNbO3,” Opt. Lett. 23, 171 (1998). [CrossRef]
10. M. E. Dearborn, K. Koch, G. T. Moore, and J. C. Diels, “Greater than 100% photon-conversion efficiency from an optical parametric oscillator with intracavity difference-frequency mixing,” Opt. Lett. 23, 759 (1998). [CrossRef]
11. M. H. Chou, K. R. Parameswaran, M. M. Fejer, and I. Brener, “Multiple-channel wavelength conversion by use of engineered quasi-phase-matching structures in LiNbO3 waveguides,” Opt. Lett. 24, 1157 (1999). [CrossRef]
12. M. A. Arbore, A. Galvanauskas, D. Harter, M. H. Chou, and M. M. Fejer, “Engineerable compression of ultrashort pulses by use of second-harmonic generation in chirped-period-poled lithium niobate,” Opt. Lett. 22, 1341 (1997). [CrossRef]
13. G. Imeshev, M. Proctor, and M. M. Fejer, “Lateral patterning of nonlinear frequency conversion with transversely varying quasi-phase-matching gratings,” Opt. Lett. 23673 (1998). [CrossRef]
14. M. J. Missey, V. Dominic, P. E. Powers, and K. L. Schepler, “Periodically poled lithium niobate monolithic nanosecond optical parametric oscillators and generators,” Opt. Lett. 24, 1227 (1999). [CrossRef]
15. P. E. Powers, T. J. Kulp, and S. E. Bisson, “Continuous tuning of a continuous-wave periodically poled lithium niobate optical parametric oscillator by use of a fan-out grating design,” Opt. Lett. 23, 159 (1998). [CrossRef]
16. W. P. Risk and G. M. Loiacono, “Periodic poling and waveguide frequency doubling in RbTiOAsO4,” Appl. Phys. Lett. 69, 331 (1996). [CrossRef]
17. R. G. Batchko, M. M. Fejer, R. L. Byer, D. Woll, R. Wallenstein, V. Y. Shur, and L. Erman, “Continuous-wave quasi-phase-matched generation of 60 mW at 465 nm by single-pass frequency doubling of a laser diode in backswitch-poled lithium niobate,” Opt. Lett. 24, 1293 (1999). [CrossRef]