1. Introduction

LiNbO₃ (LN) crystal suffers from serious photorefractive damage. This detrimental effect prevents development of new integrated optical devices. As we know, a congruent LN doped with >5 mol% MgO can effectively withstand this effect [1]. However, heavy MgO doping causes either the difficulty in growing high optical-quality single-crystal when codoped with rare-earth ions or the material inhomogeneity that prevents the fabrication of microdomain gratings with required homogeneity. It is imperative to seek other dopant with a lower threshold concentration of photorefractive damage, named C_th. Besides the well-known Mg²⁺, people has reported some other photorefractive prohibitors, such as Zn²⁺ [2], Sc³⁺ [3], Tm³⁺ [4], In³⁺ [5], Hf⁴⁺ [6], Zr⁴⁺ [7], [8] and Sn⁴⁺ [9]. Among them, the Sc³⁺ or Zr⁴⁺ doping shows a lower threshold (C_th is ~2 mol% only). The low C_th enables to improve the material homogeneity and the optical quality of crystal when codoped with other luminescent ions, such as Er³⁺. The low C_th also enables to increase the diffusivity and solid solubility of the codoped rare-earth ions. Therefore, an LN doped with > 2 mol% Sc³⁺ or Zr⁴⁺ would be a more promising substrate for an active or passive device.

Ti-diffused LiNbO₃ (Ti:LN) waveguide is a basic unit of a related device. Obviously, a simple way to fabricate a Ti:LN waveguide doped with Zr⁴⁺ or Sc³⁺ is to employ a bulk Zr⁴⁺ or Sc³⁺-doped substrate. However, such substrate is attained not as easily as a pure LN. As an alternative, the Zr⁴⁺ or Sc³⁺ can be incorporated into the crystal by diffusion method, and a Ti:Zr(Sc):LN waveguide is fabricated by either Ti⁴⁺/Zr⁴⁺ or Ti⁴⁺/Sc³⁺ co-diffusion or Ti⁴⁺ diffusion following Zr⁴⁺ (or Sc³⁺) doping. Previous study has shown that Zr⁴⁺ and Sc³⁺ diffuse slower than Ti⁴⁺ [10, 11 ]. In the viewpoint of optical-damage-resistant application, which requires that the Zr⁴⁺ or Sc³⁺ profile entirely covers the Ti⁴⁺ profile, only the two-step method is feasible. In the previous paper, we have reported the fabrication of the Ti:Sc:LN strip waveguide [11]. As the Zr⁴⁺ and Sc³⁺ have different chemical valences, ionic radii (0.7 nm for Zr⁴⁺ and 0.8 nm for Sc³⁺) and atomic weights (91 for Zr and 45 for Sc), the two ions show considerable difference in diffusion properties, which are closely related to design and fabrication of related waveguide. It is thus essential to carry out an independent study on the fabrication of a Ti:Zr:LiNbO₃ strip waveguide. In this work, we demonstrate the feasibility of using the two-step method to fabricate a Ti:LN single-mode strip waveguide doped with Zr⁴⁺. The waveguide features were characterized by end-fire coupling technique. The profiles of diffused Zr⁴⁺ and Ti⁴⁺ ions were studied by secondary ion mass spectrometry (SIMS). From the obtained Zr⁴⁺ concentration profile we conclude that the waveguide is optical-damage-resistant.

2. Experimental description

The Ti:Zr:LN waveguide was fabricated by Ti-diffusion following Zr⁴⁺-diffusion-doping.

(1) Zr⁴⁺-doping: A commercial 1-mm-thick Z-cut congruent LN plate with optical surface was used in present work. A 132 ± 2 nm thick ZrO₂ (99.99%) film was coated onto 2/3 surface part. The other 1/3 part of surface was uncoated for reference. After ZrO₂ coating, the plate was annealed in wet O₂. The diffusion temperature/duration was 1050 °C/10 h.

(2) Fabrication of Ti-diffused strip waveguide: An array of Ti strips with a width 8 μm and a thickness 100 ± 2 nm were delineated on the 1/2 part of the Zr⁴⁺-diffused surface (the other 1/2 part of Zr⁴⁺-diffused surface was not coated for reference). The separation of two adjacent strips is equally 200 μm. The Ti strips orient along the Y-axis of crystal. The Ti diffusion was carried out at 1050 °C for 10 h. To suppress Li₂O out-diffusion, the diffusion was also carried out in the wet O₂ atmosphere.

The sample surface can be divided into three parts: undoped area, Zr⁴⁺-only doped part and Zr⁴⁺/Ti⁴⁺-diffused array waveguide region. After the fabrication, the surface refractive indices at all the three parts were measured. On the surface part with waveguides, the measurements were carried out in the case that the light beam propagates along the direction perpendicular to strip waveguide axis so as to avoid the excitation of waveguide mode, which may affect the result. The refractive index was measured at the 1553 nm wavelength using a Metricon 2010 prism coupler, which has a working principle of measurement on critical angle of total reflection. Note that the index measured by this method should be the value at the crystal surface because the total reflection occurs there. For a Z-cut LN, one can choose a transverse magnetic/electric (TM/TE) polarization scheme to measure the extraordinary/ordinary refractive index.

The end-fire technique was used to observe the near-field pattern of guided mode. To facilitate the end-fire coupling, the sample endfaces were optically polished. Figure 1 shows schematic of the experiment. It was carried out by launching into the strip waveguide the polarized light emitted from a 1547 nm single-frequency laser via endface butt-coupling between a section of polarization-maintaining fiber and one waveguide. A polarization controller was used to control the polarization state. The magnified near-field image of guided mode was projected onto an infrared CCD camera through a 40 × microscope objective lens. The camera has a spectral response range 1440-1605 nm and an effective pixel number 1392 × 1040 with a pixel pitch 3.2/4.3 µm in lateral/vertical direction. To increase the spatial resolution, a lens with a larger numerical aperture of 0.65 was used.

 

Fig. 1 Schematic of near-field measurement of polarized mode guided in Ti:Zr:LN strip waveguide.

Download Full Size | PDF

The profile characteristics of diffused Zr⁴⁺ and Ti⁴⁺ ions were analyzed by SIMS. A time-of-flight secondary ion mass spectrometer (ToF SIMS V, ION-TOF GmbH) was used to analyze the surface Ti⁴⁺ profile and the depth profiles of ⁹¹Zr, ⁴⁸Ti, ⁶Li, ⁹³Nb and ¹⁶O. The surface profile was obtained by doing surface mapping with a raster size of 131 × 131 μm². For the depth analysis, a Cs⁺-beam of 30 nA at 3 keV was used to sputter a crater of 120 × 120 μm² on the waveguide surface and a pulsed bismuth ion beam (pulsed current: 1 pA, pulsed energy: 25 keV) was used to analyze the yields of secondary ions ⁶Li, ⁹³Nb, ¹⁶O, ⁹¹Zr and ⁴⁸Ti as a function of time. Positive secondary ions were detected. Ions from a central area of 10 (width) × 20 (length along waveguide axis) μm² inside the erosion crater rastered on the strip waveguide were detected. During the analysis, a low-energy pulsed electron gun was used to neutralize the positive charges and hence degrade the surface charge accumulation. For the same purpose, before the analysis a 30-nm thick Ag film was coated on the sample surface to be analyzed. The trace and depth of each erosion crater were measured by a Tencor Alpha Step 200 profilometer. The depth resolution was mainly determined by the roughness of crater and is better than 5 nm in our case.

3. Results and discussion

For a conventional Ti:LN waveguide, which is usually fabricated by diffusion of 6-10-µm -wide, ~100-nm-thick Ti-strip at 1030-1060 °C for 9 h, the surface Ti⁴⁺ concentration is around 12 × 10²⁰ ions/cm³ and the Ti⁴⁺-induced increment of n_o (n_e) is ~0.006 (0.012) at the 1.5 µm wavelength. It is unclear if the Zr⁴⁺ diffusion-doping contributes to the LN index, and if so, if the contribution is comparable to the Ti⁴⁺-induced increment. It is thus crucial to know the Zr⁴⁺ doping effect on the LN index. The effect can be determined by comparing the index values measured at Zr⁴⁺-only doped and undoped surface parts. In the earlier work, we have studied it and concluded that the effect is within the error and is small compared to above-mentioned index increment of usual Ti:LN waveguide [12]. Present work provides further evidence for this argument. The index measurement, which has an error of ± 1 × 10⁻³, shows that after Ti diffusion the n_o values are 2.2113 and 2.2110 at the Zr⁴⁺-only doped and undoped surface parts, respectively, and the n_e values are 2.1374 and 2.1375, respectively. We can see that Zr⁴⁺ contribution to the LN index is on the order of 10⁻⁴ for both cases of n_o and n_e, showing that Zr⁴⁺ doping has little effect.

Because Li out-diffusion usually accompanies ion diffusion into the LN crystal, it is essential to know the Li-compositions at the Zr⁴⁺-only doped and Ti-diffused surface parts of the studied sample. Here, the optical method of refractive index measurement was used to estimate the Li₂O-contents at the undoped, Zr⁴⁺-only doped and waveguide array surface parts of studied sample on the basis of the Li₂O-content-dependent birefringence [13], [14]. The 10⁻³ error bar of index leads to 0.1 mol% Li₂O-content uncertainty. For reference, estimation was also carried out for the as-grown plate from the measured index, n_o = 2.2112 and n_e = 2.1374. As expected, it has a Li₂O-content 48.6 ± 0.1 mol%, which is in good agreement with the nominal value. The Li₂O-contents at the Zr⁴⁺-only doped and undoped surface parts are evaluated as 48.6 and 48.5 ± 0.1 mol%, respectively. Both are identical and similar to that of congruent plate, showing that Li₂O out-diffusion is not measurable for the Zr⁴⁺-only doped and undoped surface parts.

It is impossible to accurately evaluate the Li₂O-content on the surface of an individual strip waveguide on the basis of the index measurement by prism coupler. In principle, the prism-coupling method is also applicable to a strip waveguide. There is, however, a difficulty in getting accurate results with a commercial prism coupler, which is designed for the thin film and slab waveguide rather than the strip waveguide. Because the light spot is ~1 mm in diameter while the width of a strip waveguide is typically several micrometers only (8 μm here), the measured pattern of light reflected from sample surface is mainly contributed from the substrate surface instead of the narrow waveguide surface. In addition, the Ti⁴⁺ presence induces local index increase, making it impossible to obtain the correct Li composition information of strip waveguide. One has no choice but to evaluate it indirectly and roughly based on the index data measured from the whole surface part of waveguide array. As mentioned above, two adjacent 8-μm-wide strip waveguides separate by 200 μm. Such a large spacing results in that the area part of waveguide surface seen by the light spot is < 5%. Thus, the effect of strip waveguide presence on the result should be small. Based upon this consideration, we have done the measurement on the surface part of waveguide array. The results show that the refractive indices at the surface part of array waveguide region are n_o = 2.2110 and n_e = 2.1386, yielding a Li₂O-content of 48.4 ± 0.1 mol%, which has little difference from the values at the Zr⁴⁺-only doped and undoped surface parts, 48.6 and 48.5 ± 0.1 mol%, respectively. Thus, the array waveguide surface remains still the congruent composition.

It is concluded that Li₂O out-diffusion is not serious for all the three surface parts of the sample studied here. This is possible from the point of view of charge equilibrium requirement. As we know, in the case of Ti⁴⁺ single diffusion in LN, Ti⁴⁺ enters into both Li and Nb sites. It is possible that the Li₂O out diffusion is effectively suppressed as a large number of defects [1 mol% antisites (Nb_Li) and 4 mol% Li vacancies] ensure the charge equilibrium. The situation of Ti⁴⁺/Zr⁴⁺ co-doping is similar to that of Ti⁴⁺ single diffusion because both ions have the same chemical valence.

One can further conclude from above results and discussion that the refractive index increase in the waveguide is mainly contributed from the Ti⁴⁺ dopants while not from the Zr⁴⁺ dopants or the crystal composition reduction arising from Li₂O out-diffusion.

The end-fire experiment shows that both the TE and TM modes are well guided. The waveguide is single-mode at the 1.5 μm wavelength. Figure 2(a) shows the morphology of the waveguide surface under a magnification of 1000 ×. Figure 2(b) shows the TE and TM mode patterns. Figures 2(c) and 2(d) show the TE-(red ball) and TM-mode (green balls) light intensity profiles along the width x and depth y directions of waveguide. The light intensity of the guided mode follows a Gauss function A_xexp[-2(x/W_x)²] in the x direction and a Hermite-Gauss function A_yy²exp[-2(y/W_y)²] in the y direction. The black solid lines represent the Gauss or Hermite-Gauss fits to the experimental curves. As expected, the size of TE mode (W_x × W_y = 5.2 × 4.3 μm²) is larger than that of TM mode (W_x × W_y = 4.8 × 4.2 μm²) due to the weaker guiding for this polarization.

 

Fig. 2 (a) Morphology image of 8-μm-wide Ti:Zr:LN strip waveguide; (b) Near-field patterns of TE and TM modes guided in the 8-μm-wide Ti:Zr:LN waveguide at the 1547 nm wavelength; (c), (d) TE- and TM-mode light intensity profiles along width x and depth y of 8-μm-wide Ti:Zr:LN strip waveguide.

Download Full Size | PDF

The waveguide loss was evaluated from the insertion loss measured at the 1547 nm wavelength. The fiber-to-fiber insertion loss of the 8-μm-wide, 1.5-cm-long waveguide was measured to be 6.0/6.6 dB under the TE/TM polarizations. From the obtained mode sizes, the coupling loss between the waveguide and a single-mode fiber (with a mode field diameter of 10.3 μm) is 1.9/2.0 dB for the TE/TM mode. The total reflection loss is 0.3 dB. With ignored etalon effect, the waveguide loss was conservatively evaluated as 1.3/1.5 dB/cm for the TE/TM mode. The loss is much larger than that of usual Ti:LN waveguide, which can be as low as 0.2 dB/cm [15], because of rough waveguide surface [see Fig. 2(a)] and the etalon effect, which cannot be determined accurately. Depending on the resonator length, the etalon effect gives rise to 0-2.0 dB loss in the resonator formed by the two waveguide endfaces and 0-0.5 dB loss in each air resonator formed by the endfaces of waveguide and fiber. The period of interference fringes, which is on the order of submicron, requires a measurement accuracy of at least a few tens of nanometer for both the waveguide and air resonator lengths. Such an accuracy cannot be attained. So, the above-mentioned waveguide losses are given as a result of conservative estimation.

The effective index of the TE/TM mode guided in the strip waveguide was further measured at the 1553 nm using the prism coupler. To achieve it, under the same condition we have fabricated an independent sample with a dense array of strip waveguides, which can be regarded as a quasi-planar waveguide and easily characterized using the prism coupler. The separation of two adjacent waveguides is 60 μm. One can see from Fig. 3(a) that the separation is large enough that the cross-talk does not occur. The measured mode index is given in Table 1 in comparison with substrate index.

 

Fig. 3 (a) Surface Ti⁴⁺ profile and (b) depth profiles of ⁶Li, ⁹³Nb, ¹⁶O, ⁹¹Zr and ⁴⁸Ti signals.

Download Full Size | PDF

Table 1. Substrate/mode index and surface concentration of Ti:Zr:LN strip waveguide

View Table

Figure 3(a) shows the Ti⁴⁺ profile (magenta balls) on the waveguide surface. The data can be well fitted by a sum of two error functions. The fitting expression and parameter values are indicated (W is the initial Ti-strip width). The diffusion width d_x is 9.0 ± 0.2 µm, yielding a lateral Ti⁴⁺ diffusivity 2.0 ± 0.1 μm²/h. Figure 3(b) shows the depth profiles of ⁶Li, ⁹³Nb, ¹⁶O, ⁹¹Zr and ⁴⁸Ti in the studied waveguide. One can see that both the Ti⁴⁺ and Zr⁴⁺ profiles can be well fitted by a Gaussian function, indicating that the diffusion reservoir, which is shared by Ti⁴⁺ and Zr⁴⁺, has been exhausted. The fitting expressions are indicated. The Ti⁴⁺- or Zr⁴⁺-concentration profile can be written as C_{Ti or Zr}(y) = C₀·exp[-(y/d_{Ti or Zr})²], where C₀ is the surface Ti⁴⁺ or Zr⁴⁺ concentration and d_Ti (d_Zr) is the 1/e Ti⁴⁺ (Zr⁴⁺) diffusion depth that has a value of d_Ti = 5.3 ± 0.1 μm (d_Zr = 3.6 ± 0.1 μm), which yields a Ti⁴⁺ (Zr⁴⁺) bulk diffusivity of 0.70 ± 0.02 (0.16 ± 0.01) μm²/h. Note that Zr⁴⁺ diffuses much slower than Ti⁴⁺. It is due to the differences of two ions in ionic radius (0.7 Ǻ for Zr⁴⁺ and 0.6 Ǻ for Ti⁴⁺) and atomic mass (91 g/mol for Zr and 48 g/mol for Ti).

Based upon the law of mass conservation, the C₀ is determined as (14.5 ± 0.6) × 10²⁰ [(11.9 ± 1.0) × 10²⁰] ions/cm³ for Ti⁴⁺ [Zr⁴⁺], equivalent to (7.8 ± 0.3) [(6.5 ± 0.3)] mol%.

Next, attention is paid to the optical-damage-resistant issue of the Ti:Zr:LN waveguide under study. To get an optical-damage-resistant waveguide, the waveguide layer must be completely covered by the Zr⁴⁺ profile and the Zr⁴⁺ concentration at the 1/e waveguide depth should be above the threshold of photorefractive damage, named C_th, which is around 2 mol%. The Zr⁴⁺ profile in the waveguide under study covers ~70% Ti⁴⁺ profile as d_Zr = 3.6 ± 0.1 μm and d_Ti = 5.3 ± 0.1 μm. The Zr⁴⁺ concentration at the 1/e Ti⁴⁺ concentration depth is C₀/e = 2.4 mol%, which is around C_th. It is worthy to mention that the above-mentioned former requirement should be strictly described as the Zr⁴⁺ profile entirely covers the index profile instead of the Ti⁴⁺ profile. The reason is given below. For a Ti⁴⁺-diffused waveguide fabricated on a congruent LN, the 1/e depth of index change, named d_o for the ordinary ray and d_e for the extraordinary ray, is not identical to the d_Ti because the Ti⁴⁺-induced index change ∆n_o,e and the Ti⁴⁺ concentration C_Ti follows a nonlinear relationship with a power index α_o = 0.53-0.55 [16], [17] for the ordinary ray, and a near or exact linearity with a power index α_e = 0.83 [16] or 1.0 [17] for the extraordinary ray. For the waveguide under study, the relationship should remain still because Zr⁴⁺-doping has less effect on the LN refractive index. So, to get an optical-damage-resistant waveguide, it requires d_Zr ≥ d_Ti/α_o,e ^1/2. For the studied waveguide, d_Zr should be at least d_Ti/α_o ^1/2 = 7.0 μm for the ordinary ray and d_Ti/α_e ^1/2 = 5.3 μm for the extraordinary ray. One can see that the practical Zr⁴⁺ profile (d_Zr = 3.6 μm) is definitely narrower than the required for both cases of ordinary and extraordinary rays. Instead, the nonlinearity between ∆n_o,e and C_Ti does not affect the requirement on the surface Zr⁴⁺ concentration, i. e. C ₀ ≥ eC_th.

Although the waveguide under study is not entirely covered by the Zr⁴⁺ profile, the Zr⁴⁺ profile shows that the depth where the Zr⁴⁺ concentration is above C_th can be extended from d_Zr ( = 3.6 μm) to 4.0 μm. Accordingly, the Zr⁴⁺ profile covers ~60% n_o profile and 70% n_e profile. Moreover, one can see from Fig. 2(d) that the depth range of 0.0-4.0 μm, where the Zr⁴⁺ concentration is above C_th, covers majority of field distribution for both TE and TM modes, showing that the waveguide is optical-damage-resistant.

Note that the above discussion assumes that the threshold C_th has a sharp value. Actually, it does not have a sharp value as some other factors such as temperature and light intensity may take effect on it, besides the crystal composition. So, the relevant quantification given above is an approximate consideration and is only for reference. Future work should concentrate on the optimization of fabrication parameters to improve the resistance to the optical damage. The optimization is mainly based on the above-mentioned two requirements: d_Zr ≥ d_Ti/α_o,e ^1/2 and C ₀ ≥ eC_th. By applying the law of mass conservation to the Gaussian Zr⁴⁺ profile, one can reach the following equation for the initial ZrO₂ thickness τ to be coated: τ = eπ ^0.5 d_ZrC_th/(2C_s), where C_s = ρN_A/M with ρ, N_A and M representing mass density of ZrO₂ (3.864 g/cm³), Avogadro’s number and molecular weight of the ZrO₂, respectively. On the basis of the 1/e waveguide depth d_o,e = d_Ti/α_o,e ^1/2, one can determine the d _Zr and τ. With known Zr⁴⁺ diffusivity one can further design the Zr⁴⁺ diffusion time required.

Finally, there is an issue to be clarified. As we know, a proton-exchanged LN waveguide with not very low H⁺ concentration suffers from the degradation of electro-optic (EO) coefficient. This shortcoming does not occur for the Ti:LN waveguide, verified by some already developed devices such as phase modulator, mode-locked and Q-switch Ti:Er:LN waveguide lasers. Previous study has shown that bulk Zr⁴⁺-doping has less effect on the EO properties of LN [8]. One can anticipate that the possibility is small that Zr⁴⁺ diffusion-doping affects the EO properties of Ti:LN waveguide noticeably.

4. Conclusion

We have demonstrated Ti:Zr:LN strip waveguide fabricated by Zr⁴⁺-diffusion-doping followed by Ti diffusion. It was shown that Zr⁴⁺-doping has little effect on LN index, Li₂O out-diffusion is not measurable, and index increment in the Zr⁴⁺/Ti⁴⁺-codoped waveguide layer is mainly contributed from Ti⁴⁺ dopants. Optical studies show that the waveguide well supports both the TE and TM modes, is single-mode at the 1.5μm wavelength, and has a loss less than 1.3/1.5 dB/cm for the TE/TM mode. SIMS study reveals that 60% ordinary index profile and 70% extraordinary index profile are covered by the Zr⁴⁺ profile part having the Zr⁴⁺ concentration above the threshold of optical damage, implying that the waveguide is optical-damage-resistant.