1. Introduction

Lithium niobate (LiNbO₃) single crystals have interrelated dopant-dependent optical, electrical and mechanical properties exploited in a vast number of applications [1,2,3]. Transition metals like Fe, Mn, Cu or Ni increase the photorefractive sensitivity [4], while Mg, Zn and many other metals, called optical damage resistant (ODR) dopants, if added above a threshold concentration, strongly reduce the optical damage [5]. Rare-earth (RE) ions introduced to LiNbO₃ crystals serve as laser activators [6] or active centers in coherent quantum optical experiments [7]. To understand the effect of di-, tri-, and tetravalent cations on the physical properties of the crystal their substitution sites (near Li⁺ or Nb⁵⁺ sites) and the charge compensation mechanisms have to be known. Defects induced by such dopants have been studied by various chemical and physical methods as well as theoretical modelling. Many methods reveal information only on some majority fraction of a given dopant in the mol% concentration range (e.g. Rutherford backscattering, extended X-ray absorption fine structure), leaving the site occupation for the remaining part unexplored. Several spectroscopic methods (e.g. EPR, optical spectroscopies, etc.), however, are sensitive enough to characterize the defects present in the crystal in a much lower concentration range. One of these methods is the infrared absorption spectroscopy of hydroxyl ions. $O{H^ - }$ ions are always present in as grown LiNbO₃ in the 10-100 ppm range [8]. They contribute to the charge compensation process of both intrinsic and extrinsic defects, proving at the same time to be excellent probes of the defect structure of the crystal.

In undoped, congruent LiNbO₃ (cLN), Li_1-5xNb_1+xO₃, x≈0.01, a broad $O{H^ - }$ band with a halfwidth of about ≈30 cm⁻¹ appears at about 3480-3485 cm⁻¹ consisting of at least 3–5 overlapping components due to hydroxyl ions trapped by various intrinsic defects related to non-stoichiometry [9,10]. Kong et al [11] attempted to analyze the effects of extrinsic defects on the $O{H^ - }$ spectra of low doped cLN crystals, and observed weak changes allegedly related to the contribution of $O{H^ - }$ ions located close to the dopant. They attributed the small increase of the components at 3483, 3484, and 3487 cm⁻¹ to the effect of Mg²⁺, In³⁺, and Ti⁴⁺ dopants occupying Li⁺ sites, respectively. Due to the complex shape of the band strongly depending on the thermal history of the crystal [12], which may suppress the weak effect of dopants on the $O{H^ - }$ spectrum, these results should be taken with caution. For higher dopant concentrations above the so-called photorefractive damage threshold concentration, new characteristic $O{H^ - }\; $ bands appear at wavenumbers depending mainly on the valence state of the dopant. Such bands have been observed first for Mg²⁺ [13] at about 3534 cm⁻¹, then for other divalent and trivalent ODR dopants such as Zn²⁺ at ≈3530 cm⁻¹ [14], In³⁺ at ≈3506–3508 cm⁻¹ [15,16] and Sc³⁺ at ≈3510 cm⁻¹ [17]. For the tetravalent dopants Hf⁴⁺ [18], Zr⁴⁺ [19] and Sn⁴⁺ [20], however, the “new” bands at about 3490–3500 cm⁻¹ could hardly be distinguished from the broad $O{H^ - }\; $ band of undoped cLN crystals, therefore the exact value of the involved vibrational frequency is uncertain. In the case of Mg, it was assumed that in over-threshold crystals Mg²⁺ ions may also occupy Nb⁵⁺ sites with neighboring $O{H^ - }\; $ ions responsible for the new band [21]. Later it became generally accepted that above their threshold concentrations these dopants ($M^{n + }$) partially occupy Nb⁵⁺ sites forming $M_{Nb}^{n + }$ – $O{H^ - }$ type complexes responsible for the new $O{H^ - }$ absorption bands [16–19].

Remarkable interest has also been devoted to LN crystals double-doped with Mg²⁺ above the photorefractive threshold as a first dopant and trivalent transition metal (TM) or rare-earth ions as a second dopant. In such crystals two $O{H^ - }$ absorption bands were detected: one at about 3535 cm⁻¹ related to the $Mg_{Nb}^{2 + }$ – $O{H^ - }$ complex of the type discussed above, and another one induced at about 3506 cm⁻¹ by Cr³⁺ [22], at ≈3504–3507 cm⁻¹ by Fe³⁺ [23,24], and at ≈3520 cm⁻¹ by Nd³⁺ [25] as the second dopant. The latter type was first attributed to hydroxyl ions located between the Mg²⁺ and $M^{3 + }$ ions occupying neighboring Li⁺ and Nb⁵⁺ sites, respectively [25]. Later it was pointed out by Kong et al. [26], based on their nuclear magnetic resonance measurements, that this band originates from $M_{Nb}^{n + }$ – $O{H^ - }$ type complexes without a nearest Mg neighbor, this time M indicating the second dopant. Thus, apart from the change of dopant, these complexes are similar to those found in cLN crystals single-doped by over-threshold concentrations of ODR ions.

The $O{H^ - }$ spectra have also been studied in cLN crystals double-doped with Mg²⁺ and tetravalent Ti⁴⁺ ions [27,28]. In both reports, a series of crystals were grown with a fixed above-threshold Mg concentration, co-doped with increasing amounts of Ti. When the Ti concentration exceeded another threshold of about 2 mol% an $O{H^ - }\; $ band at about 3487–3488 cm⁻¹ appeared beside or instead of the well-known $Mg_{Nb}^{2 + }$ – $O{H^ - }$ band at 3535 cm⁻¹. The same result was interpreted in different ways. Xiao et al. [27] assumed that at high concentrations Ti is present in trivalent form resulting in a neutral $Ti_{Nb}^{3 + }$ – $O{H^ - }$ – $Mg_{Li}^{2 + }$ complex. Liu et al. [28], however, concluded that Ti doping raised the Mg threshold level and the 3487 cm⁻¹ band at high Ti concentration corresponded to the well-known $O{H^ - }$ band always present in nominally pure or weakly doped cLN crystals.

The spectra of hydroxyl ions in stoichiometric LiNbO₃ (sLN), x≈0, are much better resolved. For sLN crystals a narrow hydroxyl band at 3466 cm⁻¹ with a half-width of at about 3 cm⁻¹ is dominant [29] and additional well separated $O{H^ - }$ bands can be recognized for doped crystals above a relatively low dopant threshold concentration [30]. Such additional bands have been observed not only for ODR ions, but also for RE ions, such as Nd³⁺, Er³⁺ and Yb³⁺ [31], confirming the trend: the higher the valence state of the dopant, the lower the vibrational frequency of the hydroxyl ion whereby the angle between the $O{H^ - }$ dipole and the oxygen plane perpendicular to the c-axis of the crystal also becomes smaller. The defect model $M_{Nb}^{n + }$ – $O{H^ - }$ was found to be valid for all dopants, verifying that for the lower over-threshold concentrations in sLN, at least part of the given dopant ions substitute at Nb sites.

All these results show that hydroxyl ions are excellent probes of the defect structure of LiNbO₃. This is especially true for stoichiometric crystals where due to the small amount of intrinsic defects, such as antisite niobium ions and lithium vacancies, the $O{H^ - }$ vibrational bands are relatively narrow allowing the resolution of band components related to hydroxyl ions in different environments. While the incorporation of ODR and RE ions in sLN has already been studied by $O{H^ - }$ spectroscopy, similar reports on transition metals in stoichiometric LiNbO₃ are absent. The aim of this work is to investigate systematically the absorption spectra of hydroxyl ions in a number of transition metal doped sLN crystals. The expected information on the site occupation of TM ions may be helpful for understanding their effect on optical and related properties of the crystal.

In the present work, Fourier Transform Infrared (FTIR) absorption spectroscopy was applied to a series of TM-doped (Fe³⁺, Cr³⁺, Ti⁴⁺, and Fe³⁺+Ti⁴⁺ double-doped) stoichiometric LiNbO₃ crystals. Beside the well known $O{H^ - }$ band at 3466 cm⁻¹ characteristic for the nominally pure or weakly doped sLN crystals an additional absorption band of $O{H^ - }$ ions was observed for each dopant above a threshold concentration, with the vibrational frequency and polarization behavior mainly depending on their valence state. The results prove that, similarly to other dopants in LiNbO₃ crystals, part of the transition metal ions above a threshold concentration also occupy Nb sites.

2. Experimental

A series of stoichiometric LiNbO₃ crystals were grown by the high-temperature top-seeded solution growth (HTTSSG) method using a computer-controlled resistance-heated apparatus. Crystals were grown along the z direction with a diameter of 16 mm. The pulling and rotation rates were 0.18 mm/h and 46 rpm, respectively. More detail on the growth process is given in Refs. [2,29]. The nominal concentration of the dopants added to the solution in the form of Fe₂O₃, Cr₂O₃ or TiO₂ was in the range of 0.012–0.5 mol% (see Tables 1–2). The Ti and Fe content of the crystals was determined after pressure bomb assisted acidic dissolution by means of graphite furnace atomic absorption spectrometry (GFAAS) and flame atomic absorption spectrometry (FAAS), respectively, using a ContrAA-700 (Analytik Jena AG, Jena, Germany) tandem, high-resolution continuum source AA-spectrometer. The methods of sample preparation for the analysis are described in detail in Ref. [32]. For the determination of both elements, matrix-matched calibration solutions were prepared from nominally pure LiNbO₃ crystals, in order to avoid matrix-induced interferences. A Cr₂O₃ doped sLN crystal was also grown by the Czochralski method starting from a melt characterized by [Li]/[Nb] = 1.38. Both the HTTSSG and Czochralski techniques resulted in nearly stoichiometric crystals with [Li]/[Nb] ratios close to 1 as determined from the infrared absorption spectrum described in Ref. [12].

Table 1. Concentrations of Fe and Ti dopants in the growth solution and in the sLN:M (M = Fe³⁺and/or Ti⁴⁺) crystals determined by FAAS and GFAAS, respectively, and the corresponding OH⁻ bands in the crystal characterized by FTIR spectroscopy

View Table | View all tables in this article

Table 2. Cr-doped sLN crystals grown from solution and melt

View Table | View all tables in this article

The infrared absorption spectra of at about 3–5 mm thick z-cut samples were measured in the 3000-4000 cm⁻¹ wavenumber range at room temperature by a Bruker IFS 66v/S FTIR spectrometer with 0.5 cm⁻¹ resolution. The polarization dependence of the absorption bands was measured on y-cut samples using a KRS-5 wire grid polarizer.

3. Results and discussion

The $O{H^ - }$ absorption spectra of four series of sLN crystals have been measured by FTIR spectroscopy. The first and second series each contain three crystals with increasing iron or titanium concentration, respectively (see Table 1). In the third group, four crystals double-doped with Fe and Ti can be found (see Table 1), while the fourth group is composed by three Cr-doped sLN crystals (see Table 2). Typically, the total hydroxyl ion concentration changes from sample to sample, as seen in Figs. 1–4, depending on the growth conditions in an up to now unexplored way, but in our experiments we focus only on the band positions and the relative band intensities observed in a given sample.

 

Fig. 1. IR absorption spectra of sLN crystals doped only with Fe. The arrow indicates a new absorption band appearing above a threshold Fe concentration.

Download Full Size | PDF

 

Fig. 2. IR absorption spectra of sLN crystals single-doped with Ti. The arrow indicates a new absorption band appearing above a threshold Ti concentration.

Download Full Size | PDF

 

Fig. 3. IR absorption spectra of sLN crystals double-doped with Fe and Ti. The arrows indicate the absorption bands appearing above a threshold Fe or Ti concentration.

Download Full Size | PDF

 

Fig. 4. Sample cut from the lower part of the 0.1 mol% Cr-doped HTTSSG-grown sLN crystal with a diameter of 16 mm (a), and the IR absorption spectra of the central green and the violet regions (b). The inset shows the narrow overlapping bands between 3480-3520 cm^-1.

Download Full Size | PDF

Figure 1 shows the IR absorption spectra of the stretching vibrational mode of $O{H^{ - \; }}\; $ ions in Fe doped sLN crystals. For low Fe concentration (0.06 and 0.12 mol%), only the main absorption band, characteristic for the pure or slightly doped crystals can be observed at about 3466 cm⁻¹ with a weak sideband at ≈3480 cm⁻¹. From the ratio of the band intensities R = I₃₄₈₀/I₃₄₆₅, the crystal composition is estimated for both crystals to be [Li]/[Nb] ≈ 0.998, which is very close to the perfect stoichiometric composition [12]. For higher Fe concentration (0.5 mol% in solution), the main $O{H^{ - \; }}$ band at 3466 cm⁻¹ is strongly suppressed and a new band appears at about 3502 cm⁻¹. This value reminds that observed at ≈3504 cm⁻¹ for Mg + Fe double-doped congruent crystals above the Mg-threshold [24,33].

Figure 2 shows the IR absorption spectra of the second series of sLN crystals doped only with Ti. Again, for low Ti concentration (0.012 and 0.06 mol% in solution), the main absorption band at 3466 cm⁻¹ with its sideband at 3480 cm⁻¹ can only be observed. However, for 0.12 mol% nominal Ti concentration, an additional new band has been detected at 3485 cm⁻¹. In all cases, the [Li]/[Nb] ratio estimated from the intensity ratio R = I₃₄₈₀/I₃₄₆₅ was higher than 0.996, close to the stoichiometric composition. Again, the frequency of the new band agrees quite well with those observed at ≈3487 cm⁻¹ in Mg + Ti double-doped cLN crystals for high enough concentrations for both dopants [27,28]. It is also worth mentioning that this new band at 3485 cm⁻¹ in the tetravalent Ti doped sLN crystal appears at lower dopant concentration than the 3502 cm⁻¹ band in the trivalent Fe doped sLN.

Figure 3 shows the four IR absorption spectra of the third series of sLN crystals double-doped with Fe and Ti. In the first three crystals the nominal Fe and Ti concentrations were identical, while in the fourth crystal the Fe concentration was doubled (see Table 1). Again, the R = I₃₄₈₀/I₃₄₆₅ intensity ratios clearly show that the crystal composition is nearly stoichiometric in all cases. The “new” $O{H^ - }$ absorption bands observed already in single-doped crystals appeared, however, at lower dopant concentrations: the 3485 cm⁻¹ Ti–OH band was detected for as low as 0.06 mol% nominal Ti concentration, while the 3502 cm⁻¹ Fe–OH band appeared weakly for 0.12 mol% Fe content, and became clearly seen for 0.24 mol% Fe in the host (see also Table 1). It has to be emphasized again, that in the double-doped crystals the Ti–OH band appeared at lower Ti concentration than the Fe concentration where the Fe–OH band could be detected.

The fourth series contains Cr-doped sLN crystals. The Cr dopant in LN is known to induce a green coloration. Our sLN crystal grown with 0.1 mol% Cr by the HTTSSG method looked pale green with violet regions exhibiting three-fold symmetry in the bottom part of the crystal, probably due to an instability of the incorporation of the dopant (see Fig. 4(a)). Violet coloration has also been observed in LiNbO₃:Mg,Cr double-doped crystals grown by the Czochralski method [22], and was erroneously interpreted by the presence of tetravalent chromium. The $O{H^ - }$ absorption spectra for samples cut from the upper part and the central green region of the bottom part of the crystal were practically identical showing only the main absorption band characteristic for the pure or slightly doped crystals at about 3466 cm⁻¹ with the weak sideband at ≈3480 cm⁻¹. New features could be observed, however, in the violet regions: (i) some weak, but very broad bands between 3100-3400 cm⁻¹, and (ii) several (8-10) narrow overlapping lines between 3480-3520 cm⁻¹ (see Fig. 4(b)), probably due to Cr–OH centers with different defect environments. Three Cr_Nb–H centers have also been characterized by EPR and electron nuclear double resonance (ENDOR) in LN:Cr samples taken from the inner, brown core of an otherwise green crystal boule grown from a LiNbO₃ melt containing potassium as a flux component and 1 wt.% of Cr [34]. The color comes from Cr³⁺ centers with different substitution, green and brown corresponding to the dominance of Cr_Li and Cr_Nb centers, respectively [34,35]. The violet and brown colors observed by us and by Grachev [34], respectively, are certainly different, and may be related to the different growth methods resulting in different [Li]/[Nb] ratios and different Cr concentrations in the crystals. It is clear, however, that in both cases the inhomogeneous incorporation of the Cr dopant is responsible for the unexpected coloration, which cannot be easily reproduced. Unfortunately, no OH absorption measurements have been performed on the sample of Grachev [34]. Due also to the fact that the positional information for the proton provided by the ENDOR and IR absorption measurements is complementary (azimuth for ENDOR and polar angle for IR, see below) a detailed comparison of the results is presently not feasible.

By increasing the Cr concentration to 0.5 mol%, the crystal grown by the HTTSSG method became darker green than that with 0.1 mol% Cr. The coloration of the crystal was also inhomogeneous; the central core was lighter green, but no region with violet color could be detected. Only the main $O{H^ - }\; $ absorption band characteristic for the pure or slightly doped crystals was observed at about 3466 cm⁻¹ with the weak sideband at ≈3480 cm⁻¹ (see Fig. 5). In contrary to the 0.5 mol% Fe³⁺ doped sLN crystal, no new $O{H^ - }\; $ band was detected. In the sLN:Cr³⁺ crystal grown by the Czochralski method with the same amount of Cr, however, a new band appeared at 3502 cm⁻¹, similarly to the Fe doped crystal. The $O{H^ - }\; $ vibrational frequency agrees quite well with that observed in Mg + Cr double-doped over-threshold cLN [22]. It is interesting to note that in the sample cut from the upper part of the crystal boule, the intensity of the new band is much weaker than that observed in the sample cut from the lower part of the crystal (grown along the z axis). This may be related either to the fact that during the Czochralski growth starting from melt with [Li]/[Nb] = 1.38 the composition of the crystal changes along the growth axis becoming more and more stoichiometric towards the bottom, or to the effective distribution coefficient of Cr being lower than 1 increasing the Cr concentration of the crystal in the same direction. Since the hydroxyl ion concentration in the crystal is several orders of magnitude lower than the Cr concentration, the first assumption seems to be more probable. Calculating the [Li]/[Nb] ratios from the R = I₃₄₈₀/I₃₄₆₆ values of the two samples, [Li]/[Nb] = 0.9916 and [Li]/[Nb] = 0.9924 were obtained for samples cut from the top and bottom part of the crystal, respectively. Although the difference is rather small, it supports our assumption, that the crystal becomes more stoichiometric along the growth direction during the growth process. It cannot be excluded, however, that an increasing Cr concentration along the growth direction also supports the formation of complexes of the new type.

 

Fig. 5. IR absorption spectra of 0.5 mol% Cr doped sLN crystals grown by the HTTSSG and Czochralski methods. The arrow indicates a new absorption band appearing in the crystal grown by the Czochralski method.

Download Full Size | PDF

All measurements presented above were performed on z-cut samples using unpolarized light. Polarization dependent spectra were also taken on y-cut samples in order to determine the orientation of the O–H dipole with respect to the oxygen plane perpendicular to the ferroelectric c (z) axis. Earlier measurements on $O{H^ - }\; $ defects in optical damage resistant and rare-earth doped sLN crystals above the threshold showed that the O–H dipole is tilted weakly out of the oxygen plane depending on the valence state of the dopant [30,31]. Figs. 6(a)–6(c) show the dependence of the intensity of the $O{H^ - }$ vibration on the polarization direction of the incident light with respect to the ferroelectric c axis. From the ratio of the maximum-to-minimum absorbance one can easily determine the angle Θ between the O–H dipole and the oxygen plane perpendicular to the c axis using the equation taken from Ref. [36]:

 

Fig. 6. Angular dependence of the intensity of the $\textrm{O}{\textrm{H}^ - }$ absorption band for Ti⁴⁺ (a), Fe³⁺ (b), and Cr³⁺ (c) doped sLN crystals (dots are measurement points, the solid lines are fits using the function taken from Ref. [36]).

Download Full Size | PDF

Figs. 7(a)–7(b) summarize the experimental results obtained for TM ion doped sLN as compared to those published for ODR and RE ion doped crystals [30,31]. It is clearly seen that both for vibrational frequencies and Θ angles there is an obvious dependence on the valence state of the dopants independently from their position in the periodic table of elements. This is a generalization of the trend suggested earlier: the higher the valence state of the dopant, the lower the $O{H^ - }$ vibrational frequency and the closer the O–H bond direction to the oxygen plane.

 

Fig. 7. Vibrational frequencies of $\textrm{O}{\textrm{H}^ - }$ ions in transition metal doped sLN crystals as compared to those observed in optical damage resistant and rare-earth ion doped and undoped sLN published in Refs. [30,31,29] (a). Angle Θ between the O–H dipole and the oxygen plane perpendicular to the c axis (b). The full square represents Θ calculated for the undoped sLN crystal [37].

Download Full Size | PDF

These tendencies observed for stoichiometric lithium niobate crystals have already been interpreted by the model shown in Fig. 8 where the dopant occupies a Nb site and the $O{H^ - }$ ion is trapped at a nearby oxygen site shared by oxygen octahedra containing in turn a Li vacancy and a structural vacancy in the involved stack of octahedra [30,31]. It was assumed that for di-, tri-, and tetravalent dopants the O–H dipole is tilted towards the structural vacancy, which is along the –c direction relative to the Li vacancy, therefore Θ is negative, while for the undoped crystal, where the regular Nb⁵⁺ ion can be regarded as a pentavalent dopant, the O–H is tilted in the positive direction (Θ = +4.3°) towards the Li vacancy, as determined from first principles theoretical calculations [37]. It may be assumed that the charge compensation of the $M_{Nb}^{n + }$ defect is accomplished by other nearby defects, possibly including a $M_{Li}^{n + }$ for n = 2 or 3, though, in particular, nearest neighbor $Cr_{Nb}^{3 + }$ – $Cr_{Li}^{3 + }\; $ pairs could not be detected by EPR (see [34] and references therein). It should be pointed out that the arguments of Kong et al. [26] against a similar compensation of $M_{Nb}^{n + }$ – $O{H^ - }$ by $Mg_{Li}^{2 + }$ in double-doped LN only hold if the $Mg_{Li}^{2 + }$ (or $M_{Li}^{n + }$) charge compensator is in the nearest neighborhood of the proton. Unfortunately, no calculations are available for $O{H^ - }$ centers in the presence of any kind of dopants in lithium niobate crystals.

 

Fig. 8. Schematic drawing of the $\textrm{O}{\textrm{H}^ - }$ location in Mⁿ⁺ doped sLN crystals

Download Full Size | PDF

Our present results contradict to the interpretations given for the cLN:Mg + Ti case in Refs. [27,28], and show that $M_{Nb}^{n + }$ – $O{H^ - }$ type vibrational bands characteristic for the valence state of the dopant can be observed not only in ODR and RE³⁺ ion doped sLN crystals, but can also be present in TM-doped lithium niobate above a concentration threshold also depending on the valence state of the dopant. For trivalent Fe³⁺ ions, this concentration was at about 0.2 mol%, while for the tetravalent Ti⁴⁺ ions it was lower than 0.1 mol% (see Table 1). It has to be noted, however, that these threshold values would be even lower if doped sLN crystals were grown closer to the perfect stoichiometric composition.

Figures 2–5 show that the new absorption bands appear beside the well-known $O{H^ - }\; $ bands at 3466 cm⁻¹ and its small sideband near 3480 cm⁻¹ characteristic for the undoped or slightly doped crystals. The main band at 3466 cm⁻¹ corresponds to isolated lithium vacancies while the small sideband might be related to $O{H^ - }\; $ ions associated to other intrinsic defects whose existence is promoted by the increasing presence of Nb_Li antisites [12]. This means that in sLN the dopants can simultaneously be present at both Li and Nb sites, i.e. only part of the dopants occupies Nb sites. Since the hydroxyl ion concentration is typically lower by orders of magnitude than the transition metal ion concentration in these crystals, only a small proportion of the incorporated dopants are involved in the $M_{Nb}^{n + }$ – $O{H^ - }$ type defect complexes while the others are assumed to occupy either Li or Nb sites as discussed in our previous paper on RE dopants [31].

The concept of the threshold concentration of dopants was first introduced in connection with the photorefractive damage of cLN:Mg crystals [13]. Taking into account the growing experimental body on similar threshold effects described by IR [13–31,33] and other spectroscopies (for reviews see e.g. [2,38]) in LN crystals with various other dopants and compositions, the observed “concentration threshold phenomena” acquire a general atomistic interpretation. All these effects are traced back to a change of dopant incorporation from Li to Nb site, depending also on crystal composition, and are valid not only for ODR dopants, but practically for all cation dopants investigated up to now. $O{H^ - }\; $ ions are the most convenient probes for the detection of such incorporation changes.

4. Conclusions

The presence of an $M_{Nb}^{n + }$ – $O{H^ - }$ type complex has been found in sLN crystals above a threshold concentration of transition metal ions. The tri-, or tetravalent dopants (Fe³⁺, Cr³⁺, or Ti⁴⁺) in these complexes occupy Nb⁵⁺ sites; the net negative charge attracts the proton of the hydroxyl ion resulting in a tilt of the O–H bond in the –c direction depending on the dopant’s valence state. The observed vibrational frequencies of the $O{H^ - }$ ions and their polarization dependences are in full agreement with the model previously established for ODR and RE ions [30,31] confirming its general validity. Moreover, the $O{H^ - }$ bands detected in heavily single- or double-doped congruent LN crystals can also be interpreted in the same way, i.e., by the appearance of $M_{Nb}^{n + }$ – $O{H^ - }$ type defects.