Improved Luminescence and Photocatalytic Proprieties of Sm 3+ -doped ZnO Nanoparticles Via Modied Sol-Gel Route: An Unied Experimental and DFT+U Approach

In the current study, a modied sol-gel route was followed to produce undoped and Sm 3+ doped (1, 3 and 5 mol %) nanoparticles. The study of opto-structural properties of Sm 3+ doped NPs was carried out both experimentally and theoretically. Complete dissolution of Sm 3+ ions into the ZnO lattice was obvious from XRD analysis. Morphological evolution with doping was studied using FESEM and TEM. XPS was carried out to conrm the presence of Sm 3+ on the surface of the doped NPs. Increasing dopant quantity resulted in a red shift of the NPs along with a reduction in band gap with increasing absorption in the visible range, and a minimum of 3.18 eV of optical band gap for Zn 0.97 Sm 0.03 O was found. Photoluminescence spectroscopy revealed a drop in the recombination rate of electron-hole with increasing doping till 3 mole %, followed by an increase for Zn 0.95 Sm 0.05 O. Photogenerated electron-hole pair recombination was revealed by the orange band in the luminescence spectra. Theoretical analysis was also carried out with density function theory (DFT). This work also unfolds the fundamental understanding of the structural properties of the synthesized NPs to enhance photocatalytic activity successfully. Later, photocatalytic activity for the optimum composition i.e., 3 mole percent, was assessed experimentally.


Introduction
ZnO nanoparticles (NPs) have gained a lot of popularity in recent years in applications such as UV lasers, photo detectors, gas sensors and photocatalysis which is attributed to the wide direct band gap along with large exciton binding energy [1][2][3][4]. Moreover, ZnO nanoparticles have an edge over other semiconductors in optoelectronics due to their non-toxicity, durability along with e cient absorption of the broad solar spectra [5]. Diverse morphology i.e. nanowires, nanorods, nanobelts, nanotubes, nanocapsules and nanohelices has gained considerable popularity among researchers as morphology and sizes play a pivotal role to optimize the optoelectronic properties of ZnO NPs [6,7]. Additionally, optoelectronic and photocatalytic properties of ZnO NPs can be enhanced through controlled band gap tuning [8][9][10]. Doping with rare earth elements is a promising approach to tune band gap of ZnO to make it favorable for optoelectronic applications [11][12][13][14][15][16][17]. Moreover, rare earth ions delay the recombination of photogenerated charges and thus photocatalytic performance is enhanced by active charge trapping [18].
Furthermore, rare-earth doped ZnO NPs modulate visible region emission because of their unique 4f electronic con guration effectively making them a promising candidate for multicolored LEDs [19][20][21].
Among rare-earth elements, samarium (Sm) has gained considerable attention among researchers in recent times, because Sm 3+ ions act as effective luminescent and trapping sites of the charge carriers in ZnO nanoparticles which makes them suitable for optoelectronic and photocatalytic applications [22][23][24].
Hydrothermal, sol-gel, gaseous phase deposition, successive ionic layer adsorption and reaction (SILAR) method, etc. methods are widely used to synthesize ZnO NPs [25][26][27][28][29][30]. A pivotal role is played by the synthesizing route to control the morphology, crystallinity and optoelectronic characteristics of the nanoparticles [31]. The modi ed sol-gel approach has numerous bene ts compared to other existing processes, including lower processing costs, simplicity, lower operating temperatures, and improved control over chemical homogenization of nanoparticles [32]. Here, as the authors believe, for the very rst time Sm 3+ was incorporated in ZnO NPs by this route followed by the experimental study of optostructural and photocatalytic properties of the synthesized NPs. The DFT study has also been carried out as it has proved to be instrumental in revealing the underlining correlation between structural, optoelectronic properties of rare-earth modi ed ZnO NPs [33].

Synthesis of ZnO NPs Doped with Sm3+
The chemical reagents used in this research were analytical grade. The nanoparticles were prepared by using a required amount of Zinc nitrate hexahydrate and samarium nitrate hexahydrate as precursors. To produce the solution of the salts, de-ionized water was used before the addition of the stabilizer i.e., ethylene glycol. The solution was stirred for an hour while maintaining a temperature of 80 C. The clear homogeneous solution was then supplemented with a double molar amount of citric acid (C 6 H 8 O 7 ) compared to zinc precursor. The solution was stirred vigorously for two more hours at constant temperature and speed. Meanwhile, the formation of thick gel had occurred and the gel was dried at 120℃. Powders were calcined at 500 C for a couple of hours. Finally, crushed ne powders were obtained by grinding the calcined powders.

Characterization
The crystal structure of the prepared nanoparticles was investigated by obtaining the powder X-ray diffraction patterns by Philips x-ray diffractometer [PW 3040-X'Pert PRO]. Field emission scanning electron microscopy (FE SEM: JEOL, JSM, 7600F) and transmission electron microscopy (Talos F200X TEM, Thermo Fisher Scienti c, USA) were employed to analyze particle morphology and size. SHIMADZU UV 2600 Spectrophotometer was utilized to measure absorbance and diffusion re ectance spectra, with the baseline adjusted with BaSO 4 and SHIMADZU RF-6000 Spectro-uorophotometer (excitation wavelength 200 nm) was employed to examine the photoluminescence (PL) spectra. K-Alpha thermoscienti c XPS-spectrometer was used to reveal the chemical states of the species.

Photocatalytic Degradation
The photocatalytic performance of the NPs under UV light was evaluated by the photodegradation of Rhodamine B (RhB) dye in an aqueous solution of basic medium (pH ~ 10). A solar simulator (Hamamatsu L8288, 500 W) was used to generate solar light and a UV pass lter was used to obtain UV light for the degradation [34][35][36][37]. The degradation of the dye under the illumination with the presence of the NPs was measured by the intensity of the absorbance peak of the UV-vis spectroscopy. The absorbance measurement was carried out for 180 min at a 30 min intervals. The stability and reusability test of the nanoparticles was carried out by 4 successive runs. The remaining NPs were extracted from the solution after each cycle to use as photocatalysis in the next photodegradation cycle.

Computational Analysis
All calculations presented here are based on density functional and done within the project-augmented method. For exchange-correlations, the Perdew-Burke-Ernzerhof (PBE) function was used. In the Generalized gradient approximation (GGA) potential were used for valance electrons of Zn (orbitals: 3d, 4s, 4p), O (orbitals: 2s, 2p) and Sm (orbitals: 4f, 5d) [38]. All the calculations were performed using Quantum Espresso simulation package simulation [39,40]. The Hubbard-based DFT + U correction scheme was utilized to counter the problem of underestimating band gaps of the standard DFT [41]. This type of correction is used to understand the behavior of compounds made with transition metals [42,43]. In the DFT (PBE) +U calculations for zinc orbital d, U eff = 10.8eV and for oxygen orbital p, U eff =7eV were used that computed the bandgap of 3.24 eV which is almost similar to the experimentally measured bandgap of pure ZnO. For the f orbital of Sm, we used U eff = 7eV. Rietveld re nement provides the primary structures of the pristine and Sm 3+ doped ZnO NPs. ZnO has a wurtzite hexagonal structure whose unit cell consists of 2 Zinc and 2 Oxygen atoms. However, for simulating the doped systems, supercells are essential [44,45]. For structural relaxation plane wave basis was used and pully stress was avoided by energy cutoff at 520 eV. In self-consistent eld calculations, between two successive steps the energy convergence benchmark was xed at 10-7 eV. Relaxation of atomic positions was done to bring Hellmann -Feynman forces below 0.01 eV/Å. Monkhorst-Pack of 12×12×6 scheme and σ = 0.05 sigma value of Gaussian Smearing was used to integrate the Brillouin zone to relax the unit cell. For accurate calculations, tetrahedron smearing method was used where the k-points mesh was 18×18×10. The spinpolarized method was followed for all calculations. ensures the complete incorporation of Sm 3+ into the matrix. On doping with Sm 3+ ions, crystallinity was decreased which is obvious from the reduced diffraction peak intensity. As Zn 2+ ions (R Zn 2+ =0.74 Å) were substituted by Sm 3+ ions (R Sm 3+ =1.09 Å), lattice disorder was generated leading to a decrease in crystallinity [19]. Sm 3+ doping expands the lattice cell as its radius is larger than that of Zn 2+ , and therefore, the diffraction peak of the (002) plane shifts subtly towards a lower diffraction angle [47].
For all the planes, Figure 1 shows an increase of full-width half maximum (FWHM) values, which indicates that crystallite size was decreased with increasing dopant amount [48]. Moreover, micro-stain was induced on doping which resulted in lattice disorder and stress generation [49]. Table 1 shows strain and crystallite size values derived from Williamson-Hall (W-H) equation [50][51][52]. For pristine ZnO average crystallite size was 31.2 nm whereas it was only 11.6 nm for 5 mole percent doping. On the nanoparticle's surface Zn-O-Sm formation impedes further crystal grain growth and thus, decreases the crystallite size [53]. Using Scherrer's equation, the crystallite size was again calculated and a similar trend was observed [54]. Microstrain increased when dopant was added and maximum microstrain was observed for Zn 0.95 Sm 0.05 O. Rietveld analysis was done using X'Pert HighScore Plus and the results were summarized in Table 2. Phase purity along with an increase of c/a value from 1.6024 to 1.6037 is con rmed by Rietveld analysis which further supports the low-angle peak shift.

Morphology of Nanoparticles
With increasing doping amount, a smaller-size morphology was evolved compared to pristine ZnO nanoparticles ( gure 2). Hexagonal nanoparticles for undoped ZnO were observed which is evident from gure 2a. When Sm 3+ replaces Zn 2+ in nanoparticles, it induces a positive charge, which affects the morphology [55]. Nanoparticles' size showed a decrease when dopant amount raises from 1 mole to 3 mole percent as particle growth is inhibited on doping( gure 2(b-c))[56]. An increase in the size of nanoparticles was observed for further doping and the shape of the nanoparticles become near-spherical along with agglomerated nanoparticles ( gure 2d).
The variation in particle size with doping was also clearly visible in TEM images ( gure 3). The particle size of hexagonal-shaped undoped ZnO NPs varied from 40 to 60 nm, with occasional outliers ( gure 3(ab)). In Zn 0.97 Sm 0.03 O nanoparticles, small-sized homogeneous nanoparticles ranging from 20 to 25 nm were observed ( gure 3(a-b)). The ndings are analogous to those of XRD and FESEM.

UV-vis Spectroscopy
Zn 1−x Sm x O (x = 0.01, 0.03 and 0.05) NPs showed a decrease in re ectance in UV-vis spectra which is shown in gure 4a. The light was scattered by grain boundaries of smaller-sized doped NPs which reduces re ectance [19]. Figure 4b illustrates the strong absorption threshold of pristine and doped NPs near 370 nm. The absorption edge widens as doping was increased and shifts to higher energy. Red shift and enhanced visible light absorption ability were observed with increasing concentration due to the narrower band gap. In this case, electron and hole are transferred between 4f orbital of Sm ion and ZnO valence or conduction band [57,58]. For Zn 0.95 Sm 0.05 O, absorption in the visible region was slightly decreased as further doping is responsible for surface defects. Kubelka-Munc formula was used for calculating band gaps and are listed in Table 3

Photoluminescence Spectroscopy (PL)
Photoluminescence spectra were studied for understanding the defect states and presented in gure 6.
Near band emission (NBE) is appeared as a sharp peak in the UV area as a result of excitonic nonradiative recombination [19]. Emission wavelengths were used to calculate band gap which were quite similar to the band gaps measured by the Kubelka-Muck function and is represented in Table 3. Deep level emissions (DLE) were represented by the broad orange peaks near 600 nm [19,63] . It is evident from gure 7e that two peaks of Sm 3d3/2 and Sm 3d5/2 are present at around 1110.27 eV and 1083.29 eV, respectively. These peaks con rm the presence of Sm3+ in the lattice structure [70]. The absence of other peaks of Sm further con rms that Sm is only present in the trivalent state and no secondary phase of Sm 2 O 3 is present in the lattice [71].  [45,76].
The bond length changes greatly near the doping positions as the radius of Sm 3+ ion is greater than that of Zn 2+ ion, which may be ascribed to the difference of the atomic radius and the number of valence electrons between dopants and the host atoms. The bond lengths of pristine and doped ZnO NPs are listed in Table 4. From the relaxed structure, it is revealed that the crystal structure shows some distortion after doping which is localized only around the impurity atom. The Sm-O bond length increased progressively with increasing doping. Due to this increasing bond length, the lattice constant (c/a) is also increasing with the doping concentrations of Sm 3+ which is evident from XRD analysis. As Sm 3+ ions turn out to be more closely space out for increasing doping, greater impurity-impurity contact expanded the bond length. This also affects electronic properties. From calculation, a direct bandgap of 0.72 eV is found at Γ point ( gure 11a). This is much less signi cant than the investigational value of 3.25 eV. Often standard DFT underestimates bandgap of ZnO. So effective on-site Coulomb interaction (U) (DFT + U) was used with density functional theory (DFT). Usually, standard Kohn-Sham eigenvalues exclude quasiparticle energies. So, the inadequate treatment of quasi-core electrons of a deeply correlated structure like ZnO which leads to a strong p-d hybridization. This is a very common issue for many strongly correlated systems.
The value for Ueff as 10.8 eV for zinc d orbital and 7 eV for oxygen p orbital was used. Thus, a direct band gap of 3.25 eV was calculated which is comparable to the value of 3.25 eV, which was found experimentally ( gure 11b). But the reduction in the bond length along with lattice parameters was measured. The localization of the d-state reduced the lattice constant which further drags the electron toward the core. For the rest of the calculations, these U eff values were used.
The spin-polarized DOS and band structure for different doping concentrations of Sm 3+ were calculated. The calculated projected density of states (PDOS) for the nanoparticles as a function of energy is plotted in gure 12. The valance band of undoped ZnO is created by rmly hybridized O 2p and Zn 3d orbitals.
Here, the impact from O 2p orbital is greater. From the color gradient in gure 12, it is evident that the conduction band minima (CBM) is composed of orbitals of both Zn and O atoms. There is no atomic magnetization is observed for pure ZnO as evident from the symmetric DOS. Later, magnetization was calculated as zero for this calculation. The upper valence band of doped ZnO from -2.5 eV to approximately -7.5 eV originates primarily from the O 2p and Zn 3d orbitals. As Sm 3+ has more valence electrons than Zn 2+ , with doping the system becomes more n-type. Here, fermi level moves towards the conduction band for all three doped cases. Interestingly, Sm 3+ has introduced some impurity states from Sm 4f orbital in the gap region. Sm 4f and 5d orbitals in the conduction band also contribute to a little extent in addition to Zn and O atom orbitals.
Contribution from Sm 4f and 5d orbitals near the conduction band region is visible from the PDOS where the contribution from Sm 5d orbitals is more prominent. So, the Sm 5d and Sm 4f orbitals are responsible for the change in the bandgap. Sm 5d orbitals pushed the conduction band region towards Fermi level. This makes the bandgap smaller as doping concentration increases gradually. Figure 13(

CONFLICTS OF INTEREST
There are no con icts of interest to declare.   Curve of (a) re ectance and (b) absorbance variation from UV-vis DRS spectra     Page 26/28 Figure 11 Band structure of pure ZnO using (a) DFT approximation and (b) DFT+U approximation.

Figure 12
Projected density of states for pristine and Sm3+ doped ZnO.