Morphology and bandgap modulation…

Figure 4-5 presents an SEM image of the powders; the average crystallite size is

~ 150 nm, which does not reveal a quantum size effect (the Bohr radius of excitons in ZnO ~2.34nm [36,37]).

Fig. 4-5The SEM image of the 10.9% Mg sample.

Shown in figure 4-6 are the PL and absorption spectra taken at RT of five annealed samples having volume ratios of MgO/ZnO equal to 0, 6, 10, 20, and 40%, respectively. All of the PL spectra were dominated by highly efficient NBE emission, which originates from free-exciton emission and its replicas.[5,10,11] The luminescence shows a Stokes shift to the lower-energy side of the absorption edge, which is frequently familiar in alloy semiconductors, where carriers feel different

potentials depending on the local concentration or arrangement of the substituting elements. This effect is larger in ZnO than in III－V materials, because the Bohr radius of the exciton in ZnO is smaller and the exciton is more sensitive to local inhomogeneity.[38] Moreover, the excitonic transition energy exhibits a blueshift of

~ 0.24 eV as the volume ratio of MgO/ZnO increases from 0 to 40%. Similar indications can be seen clearly in the absorption spectra, showing a blueshift from 3.33 to 3.56 eV with increasing MgO content. Namely, Mg doping causes a band-gap enlargement of ~ 240 meV.

3.0 3.1 3.2 3.3 3.4 3.5 3.6

0:1

Absorption (a.u.)

Photon energy (eV)

PL Intensity (a.u.)

0.06:1 0.2:1 MgO/ZnO

=(0.4:1)

0.1:1

Fig. 4-6 Photoluminescence (solid curves) and absorption spectra (dashed curves) of Mg Znx 1-xO alloys for different MgO/ZnO volume ratios.

Based on the formula (4.1): E(MgxZn1-xO) = E(ZnO) + 1.64*x (eV), where E(MgxZn1-xO) and E(ZnO) are the NBE emission peak positions of MgxZn1-xO and ZnO, respectively. The Mg contents in each of our five alloys, with volume ratios of

MgO/ZnO equal to 0, 6, 10, 20, and 40%, were calculated to be 0, 6, 11, 13, and 14%, respectively. Figure 4-7 illustrates the evaluated Mg content of the submicropowders as a function of the volume ratio of MgO/ZnO. The result reveals that MgO has a solubility limit of above 10% in ZnO, in accordance with the present synthesized approach. This tendency closely coincides with that reported for MgxZn1-xO fabricated by a solution-based route, in which MgO did not completely dissolve with ZnO for x > 0.1.[39] Therefore, the UV emission and absorption exhibits a progressive blueshift with initially increasing MgO content until x = 0.1 or MgO/ZnO(= 0.1:1), demonstrating that Mg²⁺ is incorporated into the ZnO lattice and occupies the lattice sites of Zn²⁺; thereafter, it tends to slowly blueshift with increasing MgO content, which implies that Mg²⁺ is no longer completely incorporated into the ZnO lattice.

0.0 0.1 0.2 0.3 0.4

0.00 0.05 0.10 0.15

Mg content (x) in the nanopowders

Volume ratio of MgO/ZnO

Fig. 4-7 The calculated Mg content in the Mg Znx 1-xO submicropowders as a function of the MgO/ZnO volume ratio.

4.2.2 Raman spectra analysis

To investigate the influence of Mg doping on the microscopic structures and the

vibrational properties, the micro-Raman spectra of the five samples were measured in a backscattering configuration with a fixed laser spot of about 2 μm², as shown in Fig.

4-8 with the inverse x-sequence compared to that in Fig. 4-6. We found the spectral peaks at 438, and 584 cm^-1 of pure ZnO submicropowders which originate from E2

(high) and E1 (LO), respectively. The assignments of the Raman peaks have been reported previously.[19-21] Furthermore, it is interesting to note that the line shape of the E (high) phonon at around 438 cm2 ⁻¹ depends considerably on the Mg incorporation.

300 400 500 600 700

14% Mg 13% Mg 11% Mg 6% Mg

Intensity (a.u.)

Raman shift (cm^-1)

ZnO E₂(high)

Fig. 4-8Micro-Raman spectra of the Mg Zn O submicropowders with various Mg contents. x 1-x

In alloy semiconductors, the phonons can be spatially confined, owing to either the potential fluctuations of the alloy disorder or finite crystalline, which gives rise to a relaxation of the q = 0 selection rule in Raman scattering.[21-23] Therefore, the spatial correlation length of phonon in alloys becomes finite. The localized phonon

mode will lead to the shift and asymmetric broadening of the Raman line shape.

Several groups [35,1,24,26]have found these phenomena in ZnO system doped with N, Mn, and Co elements.

The spatial phonon confinement could arise from APFs, as Mg²⁺ random substitution induces microscopic structural disorder in the periodic zinc atomic sublattice and breaks the translational symmetry. Besides, Islam et al. [40] and Lin et al. [41]proposed that the crystallite size distribution could affect the shifts of Raman frequencies and line shapes as well in Si and ZnO nanostructures. Therefore, they modified the Raman spectral intensity expression of the SC model by introducing a Gaussian crystallite size distribution (CSD) of an ensemble of spherical crystallites with mean crystallite size L and standard deviation σ. It can be written as o dispersion relation, Γ is the linewidth of the E (high) 2 phonon in the bulk ZnO, and

2 2

( ) 1/ 1 / 2

f q = +qσ is the characteristics of the CSD. Setting Lo and σ as adjustable parameters to fit with the experimental data, we calculated normalized Raman profiles of MgxZn1-xO submicropowders to depict the Raman line-shapes of the E2 (high) band; they are plotted in Fig. 4-9. The solid line represents the theoretical fits of the modified SC model, and the open dots are the experimental data.

The Lo values decrease with increasing Mg composition, corresponding to 17, 13.5, 12.5 nm, and 9 nm for x ~ 6, 11, 13, and 14%, respectively; nevertheless, the changes of σ and asymmetry Γa/Γb are reversed.

It is understandable that further Mg²⁺ incorporated into the ZnO lattice results in the increment of microscopic substitutional disorder, and hence enhances the standard

deviation σ and asymmetrical line shape (Γ > Γ ). In Mga b xZn1-xO alloy semiconductors, the correlation length L of Eo 2 (high) phonon can be interpreted as the average size of the localized regions [23]; thus, the phonon-extended region becomes smaller with increasing Mg content. Accordingly, the localized regions stand for microstructural geometries resulting from sublattice disorder, microcrystallite size or structural damage.[22] Similarly, the Raman spectral position and linewidth Δτ

can also be obtained by the fitting the spectral intensity using the modified spatial correlation mode, in which the E2 (high) phonon lines are slightly redshifted and remarkably broadened from 15.2 to 40 cm^-1 (full width at the half maximum, FWHM) as the Mg concentration in the MgxZn1−xO nanpowders is increased from x = 0 to

360 380 400 420 440 460 480 500 520 Raman shift (cm^-1)

x=13%

L₀=12.5nm σ=0.28

Γ_a/Γ_b=1.98

Fig. 4-9 Experimental and calculated line shapes of the E2 (high) band for Mg Znx 1-xO submicropowders with x = 6%, 11%, and 13%. The corresponding correlation length Lo, standard deviation σ and asymmetric broadening Γ /Γa b are also labeled.

0 2 4 6 8 10 12 14

Correlation Length L0 (nm)

Mg content x (%) in the nanopowders 10

Figure 4-10 reveals that the E2 (high) phonon linewidth Δτ and the correlation length Lo are strongly correlated with the Mg content and vary drastically above x ~ 0.1. As also revealed in figure 4-7, the actual dopant of Mg above x ~ 0.1 was out of proportion to the volume ratio of MgO/ZnO. Hence, instead of being incorporated into the ZnO lattice, the excess Mg²⁺ could form MgO clusters surrounding the crystalline MgZnO, which leads to a decrease in the grain size. Namely, the spherical-shape particles of ~ 150 nm, as observed in SEM, are polycrystals formed by the agglomeration of much smaller crystalline subcrystals similar to those shown in the TEM images of Cheng et al. [42]. Note that there is no significant blueshift of absorption and NBE due to the quantum size effect for crystal size larger than 7.4 nm [37]; hence, the blueshift of our MgZnO submicropowders is still dominated by the Mg incorporation. Therefore, the Lo and σ values are appropriate parameters accounting for the disorder due to APFs and GSD of MgxZn1-xO alloys.

Consequently, we would suggest that the APFs result in the change of the grain size

with Mg substitution; for x > 0.1, the grain size diminishes as a result of the random distribution of aggregated Mg²⁺ ions in ZnO crystal lattice to form MgO clusters or nanocrystals, which partition a larger MgZnO crystal (e.g., ~ 17–13.5 nm) into subcrystals (≤ 9 nm).

We have synthesized MgxZn1-xO submicropowders with various Mg contents of 0 ≤ x ≤ 0.14 by the sol-gel method. The room-temperature NBE photoluminescence and absorption spectra are shown to be tuned by ~ 0.24 eV towards the UV range upon more Mg substitution. In Raman-scattering studies, the microscopic nature of the substitutional disorder is discussed by analyzing the compositional dependence of the E2 (high) phonon mode in MgxZn1-xO alloy. It is shown that the Raman spectral broadening and asymmetry induced by the APFs can be quantitatively explained in terms of the modified SC model, considering GSD. With increasing Mg concentration, the APFs leads to a decrease in grain size, which arose from the surplus Mg²⁺ that could form MgO clusters surrounding the MgZnO crystalline.