Portal de Periódicos da CAPES

Phase Matching Achieved by Bandgap Widening in Infrared Nonlinear Optical Materials [ABa 3 Cl 2 ][Ga 5 S 10 ] (A= K, Rb, and Cs)

2020; Chinese Chemical Society; Volume: 3; Issue: 3 Linguagem: Inglês

10.31635/ccschem.020.202000268

2096-5745

Bin‐Wen Liu, Hui‐Yi Zeng, Xiao‐Ming Jiang, Guo‐Cong Guo,

Perovskite Materials and Applications

Open AccessCCS ChemistryRESEARCH ARTICLE1 Mar 2021Phase Matching Achieved by Bandgap Widening in Infrared Nonlinear Optical Materials [ABa3Cl2][Ga5S10] (A= K, Rb, and Cs) Bin-Wen Liu, Hui-Yi Zeng, Xiao-Ming Jiang and Guo-Cong Guo Bin-Wen Liu State Key Laboratory of Structural Chemistry, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350002 , Hui-Yi Zeng State Key Laboratory of Structural Chemistry, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350002 , Xiao-Ming Jiang *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] State Key Laboratory of Structural Chemistry, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350002 and Guo-Cong Guo *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] State Key Laboratory of Structural Chemistry, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350002 https://doi.org/10.31635/ccschem.020.202000268 SectionsSupplemental MaterialAboutAbstractPDF ToolsAdd to favoritesTrack Citations ShareFacebookTwitterLinked InEmail Large nonlinear optical (NLO) coefficient (dij), high laser damage threshold (LDT), and phase matching (PM) are crucial for the practical applications of IR NLO crystals. A great number of IR NLO materials with high dij have been explored, but many of them suffer from non-PM, which dramatically decreases their ultimate NLO output efficiencies. To settle this problem, we have developed a new strategy involving modulating normal dispersion via bandgap (Eg) widening to achieve PM in this work. We synthesized successfully three new wide Eg (∼4.0 eV) IR NLO salt-inclusion sulfides [ABa3Cl2][Ga5S10] (A= K, 1; Rb, 2; and Cs, 3) with PM behaviors, mainly ascribed to the PM cut-off wavelength blue shift that originated from the wide Eg values, with less contribution from birefringence enhancement. All the compounds possessed large second-harmonic-generation (SHG) efficiency (∼1 × AgGaS2) and hitherto the highest LDTs (188–200 MW/cm2, at 1.06 µm, 10 ns pulse width) for known chalcogenides with SHG efficiencies comparable to that of commercial silver gallium sulfide (AgGaS2) crystal. Download figure Download PowerPoint Introduction Infrared second-order nonlinear optical (IR NLO) crystals have broad applications, as found in optical communication, molecular spectroscopy, and laser guidance, because of tunable laser output via second-harmonic generation (SHG) and optical parametric oscillation.1–6 Notable NLO crystals used in the IR region are chalcopyrite-type crystals, such as AgGaS2 (AGS), AgGaSe2 (AGSe), and ZnGeP2 (ZGP).7–9 They possess strong NLO coefficients and wide-band transparency, which are optical merits for applications. However, they have some inherent drawbacks. For instance, AGS and AGSe have poor laser damage threshold (LDT); ZGP cannot be pumped by conventional pump lasers with wavelengths ranging from 1.0 to 2.0 µm because of strong two-photon absorptions at ∼2.0 μm. Thus, the development of new IR NLO materials with improved comprehensive properties is desirable to meet both scientific and technological demands. Several harsh prerequisites, including large SHG efficiency, wide optical bandgap Eg for high LDT, good IR transparency, and availability of growing bulk single crystals, are generally involved in searching potential for IR NLO materials.10 For IR NLO single crystal to be used practically to output laser power effectively in a device, phase-matching (PM) behavior is critical. In the past decades, considerable efforts have been exerted to discover excellent IR NLO crystals with the properties mentioned above, and thus, a high number of IR NLO materials with record-high SHG efficiency have been metal pnictide, chalcogenide, and oxyhalide systems.11–15 However, many of these materials, including K4GeP4Se12 (30 × AGSe),16γ-NaAsSe2 (75 × AGSe),17 CsZrPSe6 (15 × AGSe),18 A3Ta2AsS11 (A = K, Rb; 15 × AGSe),19 ACd4In5Se12 (A = Rb, Cs; 35-40 × AGS),20 Ba23Ga8Sb2S38 (22 × AGS),21 and Ba8Sn4S15 (10 × AGS),22 suffer from non-PM at typical pump laser wavelengths of 1.06 and 2.0 μm, which dramatically decreases their ultimate NLO output efficiency and hinders their practical applications. Theoretically, PM requires that the wavevector difference (Δk) between fundamental and double frequency lights equal to zero, that is, Δk = 0. It is often difficult to fulfill this condition because of the normal dispersion in materials, which could cause an optical path difference between fundamental and SHG light. In general, the birefringence (Δn) of a crystal can be used to compensate Δk and realize PM. Δn is determined by optical anisotropy, which, in turn, is primarily determined by lattice symmetry and structural assembly types. As it is well known, the last two structural factors are generally not easy to optimize, and only a few reports have particularly discussed it. Chen's group23 revealed that the orientation and magnitude of the dipole moment of asymmetric building units might affect the PM in CsM3Se6 (M = 0.33 Ga (or In)/0.67 Sn).23 Also, Pan's group24 discovered that the "zipper" arrangement of the PO4 tetrahedra with a large angle deviation enhanced the Δn of phosphates. More recently, Pan reported that large Δn enhancement could be obtained via alkaline-earth-metal substitution by tin (Sn).25 These methods are only valid in certain compounds or special circumstances. In addition, the quasi-PM technique has been widely applied to achieve PM by forming superlattice in some ferroelectrics, but the corresponding devices are mainly used for low-power applications.26 Alternatively, modulating normal dispersion rather than Δn might also be an effective way to achieve PM.27 The refractive index is an increasing function of frequency, and it usually increases more sharply at energies close to Eg than at energies far from it. For mid- and far-IR NLO materials, the normal dispersion in the IR region (photon energy < approximately 0.5 eV) could decrease by increasing Eg. Thus, the modulation of normal dispersion via bandgap widening could be an alternate strategy to achieve a shorter phase-matched wavelength. To verify the effectiveness of the strategy mentioned above, we selected the salt-inclusion chalcogenide system [AaXb][McQd] (A: alkali or alkali-earth metals; X: halogen; M: main group III or IV metals; Q: chalcogen)28–30 as the representative to achieve PM by bandgap widening. The salt-inclusion chalcogenides [AaXb][McQd] have attracted considerable attention due to their potential as IR NLO materials.31–33 They feature metal–chalcogenide host frameworks, where the voids are embedded with alkali or alkali-earth metal–halide guests. The host frameworks could be constructed using various NLO-active functional motifs (MQ4 tetrahedra),1,34,35 which are mainly covalent, and the guests are usually ionic. Both building blocks of hosts and guests and their assembly types could be modulated with flexibility in such a material system, which is highly beneficial for the optimization of bandgaps for PM and other IR NLO performances. Through high-temperature solid-state reactions, we obtained three new salt-inclusion sulfides [ABa3Cl2][Ga5S10] (A= K, 1; Rb, 2; Cs, 3) successfully with PM behavior. The PM behaviors of 1– 3 are mainly ascribed to the PM cut-off wavelength blueshift originating from their wide bandgaps. Besides, the three compounds possessed strong SHG efficiency (∼1.0 × AGS) and wide IR transparency ranges (0.31–12.5 μm) that almost cover the important 3–5 and 8–12 μm atmospheric transparency windows. Also, they had LDTs of 188–200 MW/cm2 (at 1.06 μm, 10 ns pulse width), which are the highest values for chalcogenides with SHG efficiencies comparable with that of commercial AGS. These high performances indicate good potential for high-power IR NLO applications. Experimental Method Syntheses The starting reactants Ba metal (99.9%), Ga metal (99.99%), S powder (99.99%), and ACl (A = K, Rb, Cs) powder (99.99%) were purchased from Aladdin Chemistry Co. Ltd. (Shanghai, China). All operations were handled in an argon atmosphere glovebox (both O2 and H2O are limited below 0.1 ppm). Three compounds 1– 3 were synthesized using a similar procedure: mixtures of Ba (64 mg), Ga (98 mg), S (90 mg), KCl (158 mg, 1), RbCl (175 mg, 2), and CsCl (197 mg, 3) were loaded into quartz tubes and evacuated to 10−4 Torr. Then the tubes were placed into a furnace, slowly heated to 950 °C, and stored at that temperature for 72 h. Finally, the tubes were cooled from 950 °C to 400 °C at 5 °C/h, before turning off the power. High yields of 95%, 92%, and 91% (based on Ba) for 1–3 and a bit of by-product β-Ga2S336 were isolated after washing the products by deionized water. Energy-dispersive analyses were performed with a Hitachi S-3500 SEM spectrometer. Bulk single crystals of the compounds 1– 3 were obtained by spontaneous nucleation and growth. A mixture of Ba (0.900 g), Ga (1.371 g), S (1.261 g), KCl (1.468 g, 1), RbCl (1.727 g, 2), and CsCl (1.929 g, 3) was placed into quartz tubes and flame-sealed under vacuum, slowly heated to 950 °C, stored for 150 h, and finally cooled down to 400 °C at 2 °C/h. Crystal structure characterization Powder X-ray diffraction (XRD) data were recorded on an automated Rigaku X-ray diffractometer (Cu Kα; Beijing, China) in the range of 5°–65° with a scan speed of 0.2°/min. The single-crystal XRD data were collected by using a Rigaku Pilatus CCD diffractometer (Mo Kα) at 293 K. The intensity data reductions were performed using the Crystal Clear software.37 The crystal structures were solved by direct methods and refined by full-matrix least-squares methods on F2 using the Siemens SHELXTL version 5 package of crystallographic software.38 The space group of I-4 was finally examined with ADDSYM/PLATON.39 Solid-state UV−Vis−NIR diffuse reflectance and infrared spectroscopy The diffuse reflectance spectra of the powdery samples were measured on a PerkinElmer LAMBDA 900 UV−vis spectrophotometer (Shanghai, China) operating from 200 to 2500 nm with BaSO4 as a 100% reflectance standard. The diffuse reflection spectra were converted to absorbance data by the Kubelka−Munk function: α/S = (1−R)2/2R (where α is the absorption coefficient, S is the scattering coefficient, and R is the reflectance).40 In addition, IR spectra ranging from 4000 to 400 cm−1 were conducted on polished single-crystal pieces by using a Nicolet Magana 750 FT-IR spectrophotometer (Shanghai Leici, China) at room temperature. SHG and LDT measurements Powder second‐harmonic generation (SHG) measurements were performed by a home-made SHG measurement system and investigated using a modified Kurtz−Perry technique with a 1.91 μm laser as fundamental frequency light. Microcrystalline samples 1– 3 were sieved into five distinct particle size ranges of 30−50, 50−75, 75−100, 100−150, and 150−200 μm, for the SHG PM measurements. IR NLO material AGS samples were used as the reference and prepared in the same procedure. The SHG signals (0.955 μm) were detected by an Andor CCD camera and software (Andor Oxford Instruments, Belfast, United Kingdom). LDT measurements were performed on the polished surface of single-crystal pieces using a 1.06 μm laser beam (pulse width τp = 10 ns, repetition rate f = 1 Hz) as the damage-inducing source. An optical concave lens was applied to modulate the diameter of the laser beam to acquire different power densities. The investigations were operated by gradually increasing the laser power densities until the sample got damaged by a color change or if a piece cracked. The damage threshold was calculated from the equation I(threshold) = E/(πr2tp), where E is the energy, r is the spot radius, and tp is the pulse width. Computational descriptions The electronic structures were calculated based on density functional theory (DFT) using the CASTEP package.41,42 The geometrically optimized supercells of all compounds were used for the calculation. The generalized gradient approximation43 was used as the exchange-correlation functional, and the norm-conserving pseudopotential44 was chosen. The plane-wave cut-off energy at 800 eV and the threshold of 10−5 eV were chosen for the self-consistent field convergence of the total electronic energy. The orbital electrons of K−3p64s1, Rb−4p65s1, Cs−5p66s1, Ba−5p66s2, Ga−3d104s24p1, S−3s23p4, Se-4s24p4, and Cl−3s23p5 were treated as valence electrons. The Brillouin zone integral was performed using 2 × 2 × 2 Monkhorst–Pack κ-point meshes. The frequency-dependent SHG susceptibilities were studied according to density functional perturbation theory (DFPT) as implemented in the ABINIT computer code package45,46 and sum formalism of Sharma.47 The contributions to SHG susceptibility could be divided into three major parts: the interband transitions χinter(2ω, ω, ω), the intraband transitions χintra(2ω, ω, ω), and the modulation of interband terms by intraband terms χmod(2ω, ω, ω). Refractive index n was obtained according to the formula: n ( ω ) = ɛ 1 2 ( ω ) + ɛ 2 2 ( ω ) + ɛ 1 ( ω ) 2 ,where ɛ1(ω) and ɛ2(ω) are respectively real and imaginary parts of the dielectric function.48 Results and Discussion Crystal structures Polycrystalline samples of 1– 3 were synthesized successfully in sealed silica tubes. Energy-dispersive spectroscopy (EDS) analysis of the single crystals confirmed the appearance of A/Ba/Ga/S/Cl elements with molar ratios of 1.0∶2.8∶5.1∶11.3∶2.2, 1.0∶2.8∶5.4∶10.3∶1.9, and 1.0∶2.9∶4.9∶11.5∶2.1 for 1– 3, respectively (see the Supporting Information Figure S1). This result agreed well with the single-crystal XRD results. The purities of the powder samples were checked by powder XRD analysis ( Supporting Information Figure S2). Bulk single crystals were grown by spontaneous nucleation, and polished pieces with millimeter-level sizes of 4.0 × 5.0 × 1.0 mm3 for 1 and 2 and 2.0 × 3.0 × 0.6 mm3 for 3 were obtained (Figure 1b). The products were stable in air or water for more than half a year. Figure 1 | Optical properties of 1–3: (a) UV−vis−NIR absorption spectra; (b) IR spectra, inset: polished pieces of crystals 1–3; (c) SHG intensities vs particle size ranges of 1–3 and AGS at 1.91 μm radiation; (d) Comparison of SHG intensities and LDT of 1–3 and AGS. Download figure Download PowerPoint The single-crystal XRD analysis indicated that 1– 3 are isostructural and crystallize in the noncentrosymmetric space group of I-4 ( Supporting Information Table S1). Their crystal structures feature a three-dimensional (3D) anionic host framework [Ga5S10]5− and a 3D cationic guest framework [ABa3Cl2]5+ (Figure 2a). Two crystallographically independent Ga atoms existed in the asymmetric unit, of which Ga(1) was tetrahedrally coordinated by four S atoms to form Ga(1)S4 tetrahedra with the distances ranging from 2.264 to 2.289 Å ( Supporting Information Table S2). Ga(2) also forms the typical Ga(2)S4 tetrahedra as the case of Ga(1), but further corner-shared with three other Ga(2)S4 tetrahedra to form T2-supertetrahedra [Ga(2)4S10]8−. Ga(1)S4 tetrahedra and [Ga(2)4S10]8− supertetrahedra shared all S apexes to form a 3D stack-like anionic network [Ga5S10]5− (Figure 2d). The 3D cationic guest frameworks were constructed by ABa3Cl tetrahedral units with Cl atoms as the centers, and further corner-shared with each other to form a [ABa3Cl2]5+ network (Figure 2c). The anionic and cationic frameworks were interlinked via ionic A/Ba-S interactions. Both the host and guest moieties form diamond-type topological structures with nodes of T2-supertetrahedral [Ga4S10]4− and tetrahedral GaS4 and (ABa3)Cl, which interpenetrated each other to create a 4-coordinated topological network with the point symbol of 66 (Figure 2b).49 Figure 2 | Overview framework (a) and topological structure (b) of 1. Frameworks of cationic guest [KBa3Cl2]5+ moiety (c) and anionic host [Ga5S10]5− moiety (d). Blue and magenta topological nodes in (b) represent T2-supertetrahedral [Ga4S10]4− and tetrahedral GaS4, respectively, and green nodes represent (ABa3)Cl tetrahedra. Download figure Download PowerPoint UV−Vis−NIR and IR transparency spectroscopy As depicted in Figure 1a, the diffuse reflectance spectra demonstrated absorption edges (bandgaps) at 3.93, 3.95, and 3.96 eV for 1– 3, respectively. These values were much larger than those of commercially used AGS (2.62 eV) and ZGP (2.10 eV).50 Wider bandgaps (up to ∼4.0 eV) decreased the dielectric damage processes when the laser was subjected to work and promote the LDT of 1– 3. IR spectroscopy was recorded in conjunction with diffuse reflectance UV−vis−NIR spectroscopy to assess the windows of optical clarity. IR transmission spectrum measurements on single-crystal pieces showed that compounds 1– 3 have the same absorption peak at 803 cm−1 (Figure 1b). Therefore, compounds 1– 3 exhibit a broad transparency region from ∼0.31 to 12.5 μm. The transparency region almost covers the critical 3–5 and 8–12 μm atmospheric transparency windows for optical applications. SHG and LDT SHG measurements were carried out with AGS as a reference by the modified Kurtz powder method.51 As shown in Figure 1c, the SHG intensities of all three compounds increased with increasing particle size at a 1.91 μm laser radiation, indicating a type-I PM behavior according to the Kurtz−Perry rule. Remarkably, compounds 1– 3 possessed high SHG intensities that were 0.95, 0.93, and 0.90 times that of AGS, respectively, at the size range of 150–200 μm ( Supporting Information Figure S3). These values were comparable to those of other wide Eg IR NLO materials, such as Na2ZnGe2S6 (0.9 × AGS),52 Pb13O6Cl9Br5 (0.9 × AGS),15 and Li2ZnSiS4 (1.1 × AGS),53 and larger than those of BaB2S4 (0.7 × AGS),54 and Ba6Zn7Ga2S16 (0.5 × AGS).55 As a PM case, the effective NLO coefficient deff was calculated using the formula deff = deff,AGS (I2ω/I2ωAGS)1/2 (I2ω and I2ωAGS are the SHG intensities for sample and AGS, respectively) with polycrystalline deff,AGS = 11.6 pm/V (polycrystalline deff,AGS is the angular average of single-crystal d36,AGS = 13.7 pm/V).51 Therefore, the experimental polycrystalline deff values are estimated to be 11.3, 11.2, and 11.0 pm/V for 1– 3, respectively. The LDT values for single-crystal pieces of 1–3 were measured by the single-pulse LDT method. As illustrated in Figure 1d and Supporting Information Table S3, the LDT values of 1– 3 are 200, 192, and 188 MW/cm2 using an incident 1.06 μm laser, showing 6.7, 6.5, and 6.3 times that of AGS (30.0 MW/cm2),50 respectively. These compounds present the highest LDTs (at 1.06 μm, 10 ns pulse width) among the documented chalcogenides with SHG efficiencies comparable to that of benchmark AGS. The process of laser-inducing damage in a crystal involves discoloration, melting, and fracture, and thus, it is highly complicated, and it could be described by two main damage mechanisms, namely dielectric and thermal damage mechanisms. The corresponding dielectric LDT is proportional to the atomic density/(n2–1), where n is the refractive index and has small values for wide-bandgap materials.56 Therefore, 1– 3 exhibited large LDT reasonably because of their wide bandgaps. Besides, the thermal expansion coefficient (TEC) is an essential intrinsic parameter that affects the thermal LDT corresponding to the thermal damage mechanism.57 Usually, NLO crystals with large TEC tend to crack under the thermal shock of laser irradiation and thus show relatively low LDT. To study the TEC of 1– 3 and the reference AGS, we analyzed the temperature dependence of their lattice parameters in situ via an X-ray diffractometer in the range of 100–400 K with a step of 20 K. As shown in Figure 3 and Supporting Information Figure S4, the linear thermal expansion (LTE) values of the a axis are 1.81 × 10−5, 1.92 × 10−5, and 1.43 × 10−5, and those of the c-axis are 0.67 × 10−5, 0.73 × 10−5, and 0.40 × 10−5 for 1, 2 and 3, respectively. These values are considerably smaller than those of AGS (2.09 × 10−5 for a axis and –1.07 × 10−5 for c axis), implying that 1–3 have higher thermal LDT than AGS. Detailed comparison of IR NLO performances (Eg, IR transparency range, SHG efficiency, and LDT) for 1–3 and typical wide-bandgap IR NLO materials is shown in Table 1. Given their wide bandgap, wide IR transparency range, large LDT, and strong SHG efficiency with PM behavior, compounds 1–3 are promising candidates for high-energy laser applications. Figure 3 | Temperature variation of the lattice parameters of 1–3 along the a axis (a) and c axis (b). The linear expansion coefficients are provided in the insets. Download figure Download PowerPoint Table 1 | Property Comparison of 1–3 and Typical Wide-Bandgap IR NLO Materials Compounds Space group Eg (eV) Transparency range (μm) deff (pm/V) LDTa (MW/cm2) Phase matchability AgGaS2 I-42d 2.62 0.48–11.4 11.6 30 PM LiGaS258,59 Pna21 4.15 0.31–11.5 5.8b ∼330 PM LiInS258,60 Pna21 3.56 0.40–12.0 7.8b 100 PM BaGa4S761 Pmn21 3.54 0.35–13.7 12.6 80 PM 1 I-4 3.93 0.31–12.5 11.3 200 PM 2 I-4 3.95 0.31–12.5 11.2 195 PM 3 I-4 3.96 0.31–12.5 11.0 188 PM aExperimental data are based on single-crystal measurements at the laser of 1.06 μm. bThese data refer to d31. Electronic structure and NLO efficiency calculations To better understand the origin of NLO properties of 1–3, we calculated their electronic structures and SHG coefficient tensors by using DFT. All these compounds possessed direct bandgaps of 3.108 eV (Figure 4a and Supporting Information Figure S5) with both maximum of the valence band (VB) and a minimum of the conduction band (CB) located at G k-point (0, 0, 0). The partial densities of states (DOS) of 1–3 are similar and shown in Figure 4c and Supporting Information Figure S6. The bottom of the CB is mostly composed of Ga-4s, Ga-4p, and S-3p states, while the top of VB originates predominately from S-3p states. Therefore, their optical absorptions could be ascribed mainly to the charge transitions from Ga−S bonding states to their antibonding states in the host [Ga5S10]5− moiety, which is responsible for the NLO effect of 1–3. The partial DOS from the guest [ABa3Cl2]5+ moieties were far from the Femi-level and made a negligible contribution to their optical absorptions. Such orbital assignments of partial DOS was confirmed further by the electron localization function pattern (Figure 4b),62 in which apparent covalent bonding could be observed between S and Ga atoms, whereas K, Rb, Cs, Ba, and Cl atoms were isolated and were mainly ionic. Although the guest [ABa3Cl2]5+ moieties did not contribute to the bandgap edges of 1–3 directly, they played a vital role as structure-guiding units in widening the bandgaps by modulating the assembly types of the host framework due to "dimension reduction" effects.63 Figure 4 | (a) Electronic band structure of 1; (b) ELF pattern of 1 at (101) plane cutting through the GaS4 tetrahedra, ELF pattern of 2 and 3 are similar to that of 1; (c) total and partial density of states of 1; (d) frequency-dependent SHG coefficient tensors of 1. The density of states and SHG coefficients of 2 and 3 are similar to those of 1 and are shown in the supporting information. Download figure Download PowerPoint Compounds 1–3 belonged to the I-4 space group and had seven nonzero SHG coefficient tensors with only two independent ones (d14, d15,) under the restriction of Kleinman's symmetry. The SHG coefficient tensors could be written as the matrix form, as follows: ( 0 0 0 0 0 0 d 15 − d 15 0 d 14 d 15 0 − d 15 d 14 0 0 0 d 14 ) As shown in Figure 4d and Supporting Information Figure S7, the calculated d14 values for 1–3 were 8.2, 8.1, and 8.0 pm/V, and their d15 values were 5.5, 5.4, and 5.3 pm/V, respectively, at the wavelength of 1.91 μm (0.65 eV). The polycrystalline deff values for 1– 3 based from the DFT calculation of dij were 8.1, 8.0, and 7.9 pm/V, respectively, which were comparable to the experimental results. PM strategy Comparison studies of PM 1–3 and those non-PM compounds in the salt-inclusion chalcogenide system were performed to verify the effectiveness of our bandgap widening strategy for PM. Taking Ba3AGa5Se10Cl2 (A = K, 4; Rb, 5; Cs, 6) with similar crystal structures as examples,28 their frequency-dependent refractive indexes were calculated based on DFT using the CASTEP package. The calculated refractive indexes (n) of 1– 6 indicated that the nz (ne) values of these compounds were larger than nx = ny (no) (Figure 5), where nx, ny, and nz were refractive indexes along the x, y, and z directions, and ne and no are refractive indexes of extraordinary and ordinary light, respectively. This result indicated that they are all positive uniaxial crystals. According to the type-I PM conditions of ne (ω) = no (2ω), where ω and 2ω refer to fundamental and second-harmonic light, respectively, the cut-off wavelengths for PM of 4–6 were 5.9, 6.4, and 5.7 μm, respectively, much longer than those of 1– 3, that is, 1.7 μm ( 1), 1.8 μm ( 2), and 1.7 μm ( 3). The larger blue shift of PM cut-off wavelengths of 1–3 than those of 4– 6 can be attributed to two reasons. First, the static birefringences of 1–3 (∼0.008) were larger than those of 4– 6 (∼0.004); second, the bandgaps of 1– 3 (3.93–3.96 eV) were considerably wider than those of 4– 6 (3.22–3.25 eV). Typically, the wider the bandgap, the flatter the normal dispersion curves in the mid- and far-IR regions, the larger energy (shorter wavelength) at which ne (ω) and no (2ω) curves cross. These results indicate that PM could be achieved by widening bandgaps. The partial contributions of bandgap widening of 1–3 to their blue shift of PM limit can be evaluated by shifting their refractive index curves to the same positions as those of 4– 6. As shown in Table 2, for all the comparison pairs ( 1, 4), ( 2, 5), and ( 3, 6), the partial contributions from the bandgap widening were significantly larger than those from birefringence enhancement. Therefore, the PM behavior of 1– 3 could be ascribed mainly to the PM cut-off wavelength blueshift originated from bandgap widening, with less contribution from birefringence enhancement. Thus, this bandgap widening strategy to achieve PM was demonstrated effectively by the salt-inclusion chalcogenides, and should also be valuable for other IR NLO material systems due to the refractive index dispersion, which is generally flat in the IR region. Figure 5 | Frequency-dependent refractive indexes of 1–3 (b, d, and f) and their analogues 4–6 (a, c, and e) along symmetry independent a and c axes. Download figure Download PowerPoint Table 2 | Partial Contributions (μm) from Eg Widening and Δn Enlargement to the PM Limit Blue Shift of 1 Compared with Analogue 4, 2 Compared with 5, and 3 Compared with 6 Comparison pair (1, 4) (2, 5) (3, 6) Eg 3.3 3.5 3.0 Δn 0.9 1.1 1.0 Total 4.2 4.6 4.0 Conclusion We have developed a strategy for achieving PM of IR NLO materials by modulating a normal dispersion via bandgap widening. The salt-inclusion chalcogenide system [AaXb][McQd] (A: alkali or alkali-earth metals; X: halogen; M: main group III or IV metals; Q: chalcogen) was studied using this strategy, and PM was achieved in three new sulfides [ABa3Cl2][Ga5S10] (A= K, 1; Rb, 2; and Cs, 3), which exhibited excellent IR NLO performances, including strong SHG efficiency (∼1 × AgGaS2) with PM behavior, large LDT (188–200 MW/cm2), and wide IR transparency region (0.31–12.5 μm). The comparison studies of PM 1– 3 and those of non-PM selenides in the inclusion chalcogenide system indicated that the PM behavior of 1– 3 was attributable mainly to the PM cut-off blueshift wavelength originating from their much wider bandgaps of ∼4.0 eV, thereby, validating the effectiveness of the strategy. This work provides an effective way to explore new PM IR NLO materials based on the non-PM ones with good SHG efficiency and LDT. Supporting Information Supporting Information is available. Conflict of Interest The authors declare no competing interests. Acknowledgments The authors wish to acknowledge Professor Ge Zhang and Dr. Bing-Xuan Li for their help in the LDT measurement. This work was supported by the NSF of China (21921001, 21827813, and 21701176), the Strategic Priority Research Program of the CAS (XDB20010100 and YJKYYQ20180006), and the NSF of Fujian Province (2018J05034). References 1. Guo S. P.; Chi Y.; Guo G. C.Recent Achievements on Middle