Improving optical and morphological properties of Mn-doped ZnO via Ar ion sputtering followed by high-temperature UHV annealing

On a silicon (Si) substrate, thin films of pure ZnO and ZnMnO (4at% Mn) were prepared using the ultrasonic spray pyrolysis technique. Zinc acetate dehydrates (ZnC₄H₆O₄_2H₂O) (99%, Sigma Aldrich) were dissolved in 100 ml of methanol at a molar concentration of 0.4 mol/l to produce ZnO. Manganese chloride tetrahydrate (MnCl2 4H2O) (Sigma Aldrich, 98%) is used as a dopant to create ZnMnO. The precursor solution contains 4% of Mn as the molar ratio concentration |Mn|/(|Mn|+|Zn|). To make sure the two solutions are homogeneous, they are each stirred for one hour at 40°C. The elaboration procedure was succeeded by spraying the precursor solutions over the heated substrates for 60 min at a temperature of 400°C.

Using a Scientaomicron UHV system with a preparation chamber and an analysis chamber, thin films of manganese-doped ZnO at 4% Mn were created. The samples were transported into the preparation or analysis chamber by a magnetically coupled probe transporter, and each chamber was pumped separately using manual or electropneumatic gate valves. When it comes to cleaning, an ion gun with a pulverized source is used to bombard the samples’ surfaces in order to get rid of different contaminants like carbon and oxygen that are on the surface layer. The cathodic sputtering ion source’s low energy is sufficient to produce flawless cleaning. It should be noted that the pressure of Ar⁺ ion gas released by a leaky valve becomes 2.10⁻⁶ mbar. With a low ion current of 3 A, the energy is fixed at 1.2 keV for 15 min. Using a standard manipulator and a pyrolytic boron nitride (PBN)-resistive device, the heating process was conducted in steps of increasing temperature (from 350°C up to 800°C) for 20 min at a time.

To study the structural properties, the EMPYREAN Diffractometer System was employed for conducting X-ray diffraction (XRD) analysis on ZnO and ZnMnO thin films. Operating at 45 kV and 40 mA, the system utilizes CuKα radiation (λ = 1.54060 Å) to capture the XRD spectrum within the angular range of 10° to 66°, employing a step size of 2θ° = 0.02°. To obtain X-ray photoelectron spectra (XPS) of the samples, a dual-anode X-ray source (DAR400) [31] was utilized. The XPS spectra were acquired by exciting the samples with monochromatic Mg-Kα (1253.6 eV) radiation. The anode voltage was set to 15 kV, and the emission current was maintained at 15 mA, corresponding to a power of 225 W. Casa-XPS software was used to calculate the atomic concentrations (at.%) of the chemical elements present on the surface. For the characterization of surface morphology and roughness, the Flex-Axiom Nanosurf atomic force microscope (AFM) was employed. Gwyddion data analysis software was utilized to process the topographic images obtained from the AFM measurements. For the purpose of investigating the presence of defects,photoluminescence (PL) measurements were carried out at room temperature using the HORIBA iHR-550 spectrometer. The measurements were conducted within the wavelength range of 340–1040 nm, and a power of 30 mW was carefully chosen as the excitation source to ensure its suitability for effectively exciting wide bandgap semiconductors.

Figure 1 exhibits the line profiles of ZnO and ZnMnO, accompanied by the 2D and 3D AFM images, denoting pure ZnO (a), untreated ZnMnO (b), and treated ZnMnO (c), respectively. The surface morphology of ZnMnO differs significantly from that of ZnO, necessitating a precisely controlled treatment procedure involving Argon ion (Ar⁺) beam sputtering [37], followed by thermal annealing [38]. The purpose of this treatment is to enhance surface roughness and diminish grain size. Notably, Figure 1c illustrates the outcome of this process. Roughness, a parameter associated with geometric defects on the actual surface, serves as an informative indicator of the adhesion and surface conditions of the deposited layers. ArzuÇolak, Zandvliet, and Poelsema [39] previously conducted comprehensive assessments of adhesion force in relation to surface morphology and power spectral density on both smooth and etched surfaces. Additionally, this technique facilitates the determination of particle size and the distance between different grain boundaries [40]. The AFM images are substantiated quantitatively in terms of Ra (average roughness), Rms (root mean square), maximum height (Z_max), and median height (Z_med), all acquired from the topographic information rendered by the atomic force microscope maps. A summary of these findings is presented in Table 1. Manganese has a significant effect on the atomic arrangement, forming small, well-resolved crystal grains on the ZnMnO surface. In addition, UHV treatment significantly improves the morphology of ZnMnO, as inferred from the mean roughness Sa calculated using Gwydion software (43.03 nm and 13.5 nm for pure ZnO and ZnMnO, respectively). Importantly, the maximum value of 288.84 observed in ZnMnO grains surpasses that of pure ZnO by threefold. This discovery holds profound significance since a rough surface optimizes the conversion of light, thereby heightening the efficiency of solar cells [41].

Fig. 1.

Figure 2 illustrates the XRD patterns obtained for the two samples. The diffraction peaks of pure ZnO and treated ZnMnO exhibit well-matched indexing with the würtzite crystal structure (space group P63mc) [JCPDF card 36-1451], with preferable orientation along (002). This suggests that the synthesized ZnO crystals are predominantly aligned along the c-axis. Moreover, it is worth mentioning that the c/a values closely resemble those of the ideal würtzite structure (ca=83)$({c \over a} = \sqrt {{8 \over 3}} )$.

Fig. 2.

Figure 2 shows fluctuating disturbance halos appearing on the diffraction spectrum above (hump) and below (hollow) the line representing the noise contained in the diffraction. The spectra of the two ZnO and ZnMnO samples are located in the two regions: 2θ (30°–32.5°) and (37.5°–47.5°), repectively. This appearance is caused, as indicated by other authors [42, 43], by amorphous materials emerging during heat treatment undergone by the two samples during the processes of their successive elaborations. Such treatment forms ultrathin layers of silicon oxide. or silicate for the as-prepared ZnO. Moreover, it can be seen that these halos become more pronounced and wider for the ZnMnO samples. This increase in dimensions is-due to the addition of an ultra-thin layer of manganese oxide. MnO or Mn₂O₃ on the surface of the manganese-doped ZnO sample.

The XRD analysis reveals that the ZnMnO crystal is free from any additional phases or impurities. Interestingly, the major peaks observed in the XRD pattern of ZnMnO show a slight shift towards lower angles compared to pure ZnO. This shift suggests a minor expansion of the unit cell in the ZnMnO sample. Consequently, the calculated lattice parameters for ZnMnO are slightly larger than those for undoped ZnO. These findings strongly suggest the successful incorporation of Mn²⁺ ions (with an ionic radius of 0.83 Å) into the Zn²⁺ lattice sites (with an ionic radius of 0.74 Å). The findings of the XRD analysis are summarized in Table 2, which provides the position peaks (2θ°), corresponding diffraction planes, and the corresponding d_hkl interplanar spacing. Notably, these results demonstrate a high level of consistency with similar XRD studies conducted on Mn-doped ZnO films synthesized through spray pyrolysis [44–49].

Moreover, the average grain size (D) is determined through the XRD spectroscopy study by calculating it as a function of the broadening of the highest intensity peak corresponding to the diffraction plane (002). This calculation utilizes Debye Scherer’s formula, as given in Eq. 1 [50]. 1 D=0.9λβ(rd)cosθ $$D = {{0.9\lambda } \over {\beta (rd)\cos \theta }}$$ Where λ is the X-ray wavelength (Cu-Kα: 1.5406 A°), β (rd) is the full width-half-maximum (FWHM), and θ (rd) is the peak position. The lattice parameters (a, c, and c/a) were calculated using formula (2): 2 1dhkl2=43(h2+hk+k2a2)+l2c2 $${1 \over {d_{hkl}^2}} = {4 \over 3}({{{h^2} + hk + {k^2}} \over {{a^2}}}) + {{{l^2}} \over {{c^2}}}$$ Where d_hkl is the interplanar spacing determined by Bragg’s law and h, k, and l are the Miller indices. the strain (ε) in the film is calculated using the following formula: 3 ε=β(rd)cosθ4 $$\varepsilon = {{\beta (rd)\cos \theta } \over 4}$$

The particle size of pure ZnO measures 25.20 nm, whereas the crystallite size of ZnMnO demonstrates a significant reduction to 12.41 nm. This decrease can be attributed to the enhanced solubility of Mn in ZnO, as observed in the study by Staumal et al. [51]. Additionally, it is important to note that the material experiences an increase in strain. This change arises from multiple factors, including the exchange of ions with different ionic radii, shifts in lattice constants, the introduction of impurities or defects, and the thermal extinction coefficient between the substrate and the film [52].

Figure 3 depicts the variance in β_T cos(θ) vs. 4Sin(θ) (Williamson–Hall analysis) [53]. Equation (4) depicts (in linear form) y=mx + c, where m = strain and c = Kλ/D, such that the slope of the linear plot of cos(θ) versus Sin(θ) is lattice strain ε, and the intercept is Kλ/D. 4 βTcosθ=ε(4sinθ)+KλD $${\beta _T}\cos \theta = \varepsilon (4\sin \theta ) + {{K\lambda } \over D}$$

Fig. 3.

The W-H analysis for grain size look comparable; however, the results for strain appear to be widely off. These findings are consistent with the results obtained from the AFM study, providing further support for the notion that the diminished crystallite size predominantly stems from the distortion of the ZnO host lattice caused by the presence of Mn impurities. Consequently, this distortion facilitates accelerated crystal growth and the amalgamation of crystal grains.

This study utilizes X-ray photoelectron spectra (XPS) to investigate the impact of Mn incorporation and UHV treatment on the chemical composition of ZnO thin films. Representative XPS results of clean ZnO and ZnMnO(untreated, sputtered, and annealed) are showcased in Figure 4(a). Notably, all photoelectron peaks (Zn 2p, Mn 2p, O 1s, C 1s, Zn 3s, Zn 3p, and Zn 3d) and Auger transition peaks (C-KLL, O KLL, and Zn LMM) associated with ZnO or ZnMnO materials are accurately identified. For data correction, the C 1s peak, located at 284.6 eV, is utilized as a reference. The atomic percentages (at%) of oxygen, manganese, and zinc elements are derived from the O 1s, Mn 2p_3/2, and Zn 2p_3/2 signals and are presented in Table 3. Figure 4(b) reveals a noteworthy observation that the C1s carbon signal is more pronounced in ZnMnO-compared to clean ZnO. Furthermore, the untreated ZnMnO exhibits a more significant C1s carbon signal than the annealed ZnMnO.

Fig. 4.

In Figure 4(c), the presence of the Zn 2p core level, which represents the divalent state of zinc ions in ZnO, is confirmed. The Zn 2p_3/2 (1021.90 eV) and Zn 2p_1/2 (1044.91 eV) photoelectron peaks exhibit a double peak energy separation of 23.01 eV, consistent with previous studies [54, 55]. However, when 4% Mn doping is introduced, the ZnMnO signals of Zn 2p_3/2 and Zn 2p_1/2 become less resolved due to the presence of a carbon contamination layer on the surface [56, 57]. Notably, following UHV treatment of ZnMnO, there is a substantial increase in the intensity of the Zn 2p signals, accompanied by a shift towards lower binding energies (approximately 0.45 eV) [58]. This observed shift could be attributed to morphological changes, as illustrated by the AFM images.

Figure 4(e) presents the high-resolution Mn 2p spectra, revealing prominent peaks at binding energy values of 641.8 eV and 653.9 eV, which correspond to Mn 2p_3/2 and Mn 2p_1/2, respectively. Interestingly, the Mn 2p signals exhibit negligible intensity in the contaminated state. However, upon subjecting the samples to UHV treatment, a notable improvement in peak shapes was observed.

Gaussian deconvolution was employed to explore the distribution of the Mn element within the thin films. As shown in Figure 5(a), the presence of distinct peaks at 641.10 eV and 653.0 eV provides unequivocal evidence for the existence of Mn²⁺ ions. These findings not only corroborate the Mn²⁺ oxidation state but also affirm the successful substitution of Mn²⁺ ions into the Zn site positions within the ZnO lattice. Furthermore, higher binding energy peaks at 642.55 eV and 654.60 eV are unequivocally assigned to Mn³⁺ ions present in the ZnMnO thin film. The coexistence of two charge states, Mn⁺² and Mn⁺³, in Mn-doped ZnO has also been reported in previous studies [59, 60], further validating the observed Mn³⁺ peaks. Additionally, the presence of two satellite peaks, located around 645.32 eV and 657.15 eV, can be attributed to the electron shake-up of Mn²⁺.

Fig. 5.

Figure 4(d) presents the high-resolution oxygen 1s signal, showcasing a binding energy of 530.65 eV for the annealed ZnMnO sample. Notably, the O 1s peak undergoes asymmetry upon Mn doping but regains its symmetrical shape following UHV treatment. Deconvolution of the O 1s peak in untreated ZnMnO, as illustrated in Figure 5(b), reveals three distinct Gaussian peaks labeled as a, b, and c. The low binding energy peak O_a, located at 530.89 eV, corresponds to O²⁻ ions participating in Zn-O and/or Mn-O bonds. The intermediate binding energy peak O_b, observed at 532.53 eV, is associated with Zn-OH species. Additionally, the high binding energy peak O_c, appearing at 533.75 eV, is attributed to chemisorbed oxygen contributing to the formation of H₂O, CO₂, and/or oxygen defects such as V_o or O_i species, as previously reported [61, 62]. Upon the removal of contaminated layers containing OH, COx, and H₂O species through Ar ion sputtering and high-temperature annealing treatment, the Gaussian fit of the O 1s peak reveals only two sub-peaks, O_a and O_b, positioned at 530.37 eV and 531.48 eV, respectively [10, 14], showing considerable reduction in Vo or/and Zn_i concentrations.