Enhanced Ultraviolet Damage Resistance in Magnesium Doped Lithium Niobate Crystals through Zirconium Co-Doping

MgO-doped LiNbO3 (LN:Mg) is famous for its high resistance to optical damage, but this phenomenon only occurs in visible and infrared regions, and its photorefraction is not decreased but enhanced in ultraviolet region. Here we investigated a series of ZrO2 co-doped LN:Mg (LN:Mg,Zr) regarding their ultraviolet photorefractive properties. The optical damage resistance experiment indicated that the resistance against ultraviolet damage of LN:Mg was significantly enhanced with increased ZrO2 doping concentration. Moreover, first-principles calculations manifested that the enhancement of ultraviolet damage resistance for LN:Mg,Zr was mainly determined by both the increased band gap and the reduced ultraviolet photorefractive center O2−/−. So, LN:Mg,Zr crystals would become an excellent candidate for ultraviolet nonlinear optical material.


Introduction
Lithium niobate (LiNbO 3 , LN), also known as the silicon of nonlinear optics, is famous for its excellent physical properties such as acousto-optic, electro-optic, pyro-electric, piezoelectric, and nonlinear properties [1,2]. Although it has been grown for more than half a century, LiNbO 3 has remained one of the hot issues for researchers up to now [3]. In LiNbO 3 , the optically-induced damage, also called the photorefraction, can be produced easily with a laser beam of several milliwatts, which seriously hinders its applications such as optical waveguides, frequency convertors, and Q-switches at high light intensities [4]. The good news is that doping MgO into LiNbO 3 solves this problem well, and the optical damage resistance for LiNbO 3 can be increased by two orders of magnitude, substantially promoting its practical use in nonlinear optics [5,6]. Nowadays, MgO doped LN (LN:Mg) still garners extensive research interest in optical waveguide, domain engineering, and quasi-phase-matching, however, it should be noted that most of these researches have only been carried out in visible and infrared regions, but not in ultraviolet (UV) [7,8]. In fact, LN:Mg exposes the visible enhancement of UV photorefraction [9], and this is a serious problem that has become a noteworthy impediment to its extension of applications into UV region.
As we know, the chemical element Zr is considered as a unique and promising dopant in LiNbO 3 since it has a lower doping threshold (~2.0 mol.%) and a optimal distribution coefficient close to one, especially the strong optical damage resistance from UV to visible [10][11][12][13]. It should be pointed out that it has been the only doped LN crystal possessing superior resistance against UV damage so far [14]. Generally, this property is thought to be related to the replacement of Zr 4+ ions against Nb 5+ on Li sites because the element Zr is adjacent to Nb in the periodic table. It has been reported that Zr 4+ ions co-dope with some transition metal elements in LN crystals [15,16], such as Fe, Mn, Cu, and Ce, which can considerably reduce the light-induced scattering as well as improve the response speed. Besides, Zr doping also improves the near infrared emission efficiency in LN:Er, LN:Yb,Er crystals [17]; meanwhile, it can also stabilize the signal output of the Ti-diffused LN waveguides under the high-power pumping without optical damage observed [18]. Furthermore, we have reported that some additional ZrO 2 doping into LN:Mg not only can finely adjust the phase-matching temperature but also realize the room temperature noncritical phase-matching [19]. The above research results suggest that Zr doping have such a strong impact on the properties of LN, so further research on its influence on the UV photorefraction of LN:Mg is very necessary and meaningful to promote the laser frequency conversion to UV wavelength region. Therefore, in this work we mainly investigated the resistance against UV damage of ZrO 2 co-doped LN:Mg crystals. Moreover, the first-principles calculations were employed to explore the effect of the crystal defect on the UV photorefractive properties in the doubly-doped LiNbO 3 .
The transmitted beam spot distortion method [20] was performed directly and evaluated the optical damage resistance of crystals in the UV region. An e-polarized beam operating at 355 nm from a continuous wave frequency-tripled solid-state laser with a maximum output power of 100 mW was focused into the 3.0-mm-thick y-plates. The laser beam was polarized parallel to the c-axis of the plates, and the highest intensity at the focal point was 2.2 × 10 5 W/cm 2 . When a crystal suffers from the optical damage, the transmitted beam spot is smeared and elongated along the c-axis with decreased intensities at the central part. Moreover, in order to further characterize the optical damage resistance at 355 nm quantitatively, we also used the two-wave coupling holographic recording method [21], namely, two coherent e-polarized beams with irradiation intensities of 400 mW/cm 2 were focused into the y-oriented plates at the intersecting angle of 28 • (2θ) to measure the light-induced refractive index changes (∆n) of these crystals.
In addition, to explore the linking mechanism between defect structure and optical damage resistance properties, the theoretical calculation for the defects and electronic structure of the crystals was performed by the Vienna Ab-initio Simulation Package (VASP) and general gradient approximation (GGA) [22,23]. The cutoff energy of the projector augmentation wave (PAW) pseudo-potentials was 400 eV with an allowed error in energy from relaxation of 10 −5 eV. Moreover, the calculations of the bulk defects in perfect LN were carried out in the super-cell containing 2 × 2 × 1 conventional unit cells with 120 atoms, meanwhile the employed 2 × 2 × 1 K-points mesh over the Brillouin-zone was generated by the Monkhost-Pack scheme [24].  We also calculated the refractive index changes by two-wave coupling holographic recording method, and the relevant results are shown in Figure 2. From this figure, it reveals that the changes of the refractive index of LN:Mg5.0 have lowered about an order of magnitude by co-doping with ZrO2, which can even be compared with that of LN:Zr2.0. Considering the inverse correlation in the change of refractive index with the optical damage resistance of the crystals, we conclude that the additional ZrO2 dopants greatly reduce the UV optical damage of LN:Mg5.0. However, it does not show a linear pattern between the refractive index changes and ZrO2 co-doping with increasing ZrO2 doping concentration in doubly crystals, and this particularity probably signifies that ZrO2 doping has had a profound effect on the crystal structure of LN:Mg. So, we utilized the first-principles calculations to find out more about the inherent mechanism of enhanced UV optical damage resistance of LN:Mg by ZrO2 co-doping.

Defects Structure in the LN:Mg,Zr
Generally, the formation energy is treated as the stability criterion of defects in materials [25]. According to the optimization model of prefect LN [26], we firstly calculated the formation energies of point defects Mg Li + , Mg Nb 3-, Zr Li 3+ , Zr Nb in LN:Mg,Zr to investigate the substitution sequence of doping ions, and the results are shown in Figure 3. It should be noted that the concentration increase of doping ions will cause the raise of Fermi level [27], so it can be observed from the figure that Mg Li + and Zr Li 3+ transform into Mg Nb

3-
and Zr Nb at the points when EF = 1.9 eV and EF = 1.6 eV, respectively, which illustrates We also calculated the refractive index changes by two-wave coupling holographic recording method, and the relevant results are shown in Figure 2. From this figure, it reveals that the changes of the refractive index of LN:Mg 5.0 have lowered about an order of magnitude by co-doping with ZrO 2 , which can even be compared with that of LN:Zr 2.0 . Considering the inverse correlation in the change of refractive index with the optical damage resistance of the crystals, we conclude that the additional ZrO 2 dopants greatly reduce the UV optical damage of LN:Mg 5.0 . However, it does not show a linear pattern between the refractive index changes and ZrO 2 co-doping with increasing ZrO 2 doping concentration in doubly crystals, and this particularity probably signifies that ZrO 2 doping has had a profound effect on the crystal structure of LN:Mg. So, we utilized the firstprinciples calculations to find out more about the inherent mechanism of enhanced UV optical damage resistance of LN:Mg by ZrO 2 co-doping.  We also calculated the refractive index changes by two-wave coupling holographic recording method, and the relevant results are shown in Figure 2. From this figure, it reveals that the changes of the refractive index of LN:Mg5.0 have lowered about an order of magnitude by co-doping with ZrO2, which can even be compared with that of LN:Zr2.0. Considering the inverse correlation in the change of refractive index with the optical damage resistance of the crystals, we conclude that the additional ZrO2 dopants greatly reduce the UV optical damage of LN:Mg5.0. However, it does not show a linear pattern between the refractive index changes and ZrO2 co-doping with increasing ZrO2 doping concentration in doubly crystals, and this particularity probably signifies that ZrO2 doping has had a profound effect on the crystal structure of LN:Mg. So, we utilized the first-principles calculations to find out more about the inherent mechanism of enhanced UV optical damage resistance of LN:Mg by ZrO2 co-doping.

Defects Structure in the LN:Mg,Zr
Generally, the formation energy is treated as the stability criterion of defects in materials [25]. According to the optimization model of prefect LN [26], we firstly calculated the formation energies of point defects Mg Li + , Mg Nb 3-, Zr Li 3+ , Zr Nb in LN:Mg,Zr to investigate the substitution sequence of doping ions, and the results are shown in Figure 3. It should be noted that the concentration increase of doping ions will cause the raise of Fermi level [27], so it can be observed from the figure that Mg Li + and Zr Li 3+ transform into Mg Nb

3-
and Zr Nb at the points when EF = 1.9 eV and EF = 1.6 eV, respectively, which illustrates

Defects Structure in the LN:Mg,Zr
Generally, the formation energy is treated as the stability criterion of defects in materials [25]. According to the optimization model of prefect LN [26], we firstly calculated the formation energies of point defects Mg + Li , Mg 3− Nb , Zr 3+ Li , Zr − Nb in LN:Mg,Zr to investigate the substitution sequence of doping ions, and the results are shown in Figure 3. It should be noted that the concentration increase of doping ions will cause the raise of Fermi level [27], so it can be observed from the figure that Mg + Li and Zr 3+  3. Zr 4+ ions only substitute Nb-sites while Mg 2+ ions occupy both Li-and Nb-sites, labelled as Zr Nb -+4Mg Li + +Mg Nb 3-.

Charge States of LN:Mg,Zr
The partial density of states (PDOS) of LN:Mg,Zr was calculated with the aforementioned optimal Mg Li + +Zr Nb model, as shown in Figure 4a. Moreover, according to the 3Mg Li + +Mg Nb 3defect model [26], the calculated PDOS of LN:Mg is described in Figure 4b for comparison. The PDOS clearly showed the DOS of each atom in the LN:Mg,Zr crystals, and the highest occupied valence band exhibits mainly O-2p features, whereas the lowest unoccupied conduction band mainly consists of Nb-4d electrons, meaning that the band gap of LN is determined by the transition energy of an electron transition from O 2− 2pstate to Nb 5+ 4d-state, consistent with the widely accepted results [28]. In our previous report [19] we employed the OHabsorption spectra and UV-visible absorption spectra to analyze the defects of these doubly co-doped crystals, and the results indicated that the Mg + Li + Zr − Nb defect clusters exist in crystals. Hence, according to the experimental results and the above calculated results of point defects, three kinds of defect clusters were constructed to find the energetically preferable configuration in LN:Mg,Zr crystals:  Table 1 shows the calculated defect formation energies, and the Mg + Li + Zr − Nb defect cluster had the lowest defect formation energy, reflecting that the existence of Mg + Li + Zr − Nb defect cluster in LN:Mg,Zr was reasonable, which is in agreement with that of our previous experimental results.

Charge States of LN:Mg,Zr
The partial density of states (PDOS) of LN:Mg,Zr was calculated with the aforementioned optimal Mg + Li + Zr − Nb model, as shown in Figure 4a. Moreover, according to the 3Mg + Li + Mg 3− Nb defect model [26], the calculated PDOS of LN:Mg is described in Figure 4b for comparison. The PDOS clearly showed the DOS of each atom in the LN:Mg,Zr crystals, and the highest occupied valence band exhibits mainly O-2p features, whereas the lowest unoccupied conduction band mainly consists of Nb-4d electrons, meaning that the band gap of LN is determined by the transition energy of an electron transition from O 2− 2p-state to Nb 5+ 4d-state, consistent with the widely accepted results [28].
eV, which is significantly greater than that of LN:Mg (2.942 eV), and the increase in band gap of LN:Mg,Zr makes charge carriers excitation from O 2-/-more difficult, which powerfully proves the above analysis and further confirms the theoretical calculation corresponding to the experiment results. Based on the analyses above, we state that the reduction in the number of UV photorefractive canters O 2-/-and the increased band gap together result in the high resistance against UV optical damage of LN:Mg,Zr.

Conclusions
In summary, we have investigated the resistance against ultraviolet damage and electronic structures of LN:Mg co-doping with ZrO2 by experiment and first-principles calculations. It is found that ZrO2 co-doping greatly improves the ultraviolet damage resistance of LN:Mg. Meantime, first-principles calculations elucidated that the enhancement of resistance against UV damage of LN:Mg by co-doping ZrO2 is caused by both the increased band gap and the reduced ultraviolet photorefractive center O 2-/-. This knowledge may open avenues to expand the nonlinear optical applications of LN:Mg to the UV region, and it can be an important candidate material for UV frequency conversion.   Moreover, it can also be seen that the density of states of O-2p and Nd-4d in LN:Mg,Zr exhibit a distinctly decreased DOS at valence band maximum (VBM) and conduction band minimum (CBM), respectively, which elucidates the reduction of charge carriers in crystals [29]. The enhancement of resistance against UV optical damage of LN:Zr is considered as the reduction of UV photorefractive centers such as O 2−/− near Zr − Nb [11,30]. It knows by the calculated DOS that the charge density of O-2p of LN:Mg is reduced dramatically by ZrO 2 co-doping, which signifies that the UV photorefractive center O 2−/− of LN:Mg,Zr is sharply decreased. Moreover, the band gap of LN:Mg,Zr is calculated to be about 3.116 eV, which is significantly greater than that of LN:Mg (2.942 eV), and the increase in band gap of LN:Mg,Zr makes charge carriers excitation from O 2−/− more difficult, which powerfully proves the above analysis and further confirms the theoretical calculation corresponding to the experiment results. Based on the analyses above, we state that the reduction in the number of UV photorefractive canters O 2−/− and the increased band gap together result in the high resistance against UV optical damage of LN:Mg,Zr.

Conclusions
In summary, we have investigated the resistance against ultraviolet damage and electronic structures of LN:Mg co-doping with ZrO 2 by experiment and first-principles calculations. It is found that ZrO 2 co-doping greatly improves the ultraviolet damage resistance of LN:Mg. Meantime, first-principles calculations elucidate that the enhancement of resistance against UV damage of LN:Mg by co-doping ZrO 2 is caused by both the increased band gap and the reduced ultraviolet photorefractive center O 2−/− . This knowledge may open avenues to expand the nonlinear optical applications of LN:Mg to the UV region, and it can be an important candidate material for UV frequency conversion.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available from the corresponding author on reasonable request.