1. Introduction

The field of nanocrystalline materials has been widely investigated in recent years for the improvement of magnetic, chemical and mechanical properties. The diluted magnetic semiconductor (DMS) has attracted the attention of researchers; a fraction of the host cations can be substitutionally replaced by magnetic ions^{[1, 2]}. This class of semiconductors has potential use of both charge and spin of electrons for spintronic devices such as spin-valve transistors, spin light-emitting diodes, logic devices and nonvolatile memory. In this regard, ZnO has become a significant DMS material for its 3.37 eV wide direct band gap at room temperature and high chemical and thermal stability^{[3, 4]}. Hence, significant research has been conducted to develop highly oriented ZnO thin films to expand their potential optoelectronic applications in various fields like transparent electrode in display, window layers in solar cells, field emitters, ultraviolet laser emission, photodetectors, piezoelectricity, bio-sensors and short wavelength LED.

Transition Metals (TM) like Fe, Co, or Ni doped ZnO have a technological advantage for their favorable magnetic, optical and electronic properties required for spintronic materials and optoelectronic devices^[2]. Among these, Fe doping can considerably insist intrinsic ferromagnetism in diluted magnetic semiconductors e.g. ZnO. The issues like microstructure of host ZnO and valence state of Fe are determinant factors of many of these properties^{[5, 16]}. Furthermore, 3d transition metals have partially filled d states. These partially filled d states contain unpaired electrons and are responsible for exhibiting magnetic behavior. The addition of 3d transition metal ions can change the Fermi energy state by raising the valance band maximum and lowering the conduction band minimum, thus reducing the band gap. Thus, it is expected that the doping of Fe metal ion into the ZnO matrix will change the optoelectronic properties as well.

For Fe-doped ZnO materials, most of the research focus mainly concentrated on the ferromagnetic behavior of Fe-doped ZnO thin films^[6-9].However, Fe-doped ZnO thin film can be an important material for utilizing its structural, electrical and optical properties in different applications; thus it is important to study these properties and the correlation between them. Furthermore, certain effects of either some dopants or preparation procedure remain unclear on Fe-doped ZnO and need to be fully investigated. Fabrication technique, dopant sources, deposition parameter, ambient environment -all these have significant impact on the prepared thin film surface and need to be rigorously studied.

Many fabrication techniques e.g. plasma enhanced chemical vapor disposition, pulsed laser disposition, sputtering, sol-gel and spray pyrolysis are found in literature for fabrication of ZnO thin films^{[1, 3, 10-17]}. Among these, thermal spray pyrolysis has several advantages including the possibility of depositing large surface films under atmospheric conditions on various substrates and from low priced chemicals^{[2, 10-12]}.Therefore, spray pyrolysis is considered as a promising technique for the cost-efficient fabrication of ZnO thin films in future. In literature, fewer groups have prepared Fe-doped ZnO by spray pyrolysis compared to other deposition methods and further studied their properties. Hence, in this work, we prepared Fe-doped ZnO thin films on glass-substrate using thermal spray pyrolysis for different Fe-doping concentrations. Structural and optical properties of the film have been investigated and the effect of doping and valence state of the Fe have been analyzed to explain the observed phenomena.

2. Experimental details

Zinc acetate (Zn(COOCH₃)₂$\cdot$2H₂O), water (H₂O), ferric chloride (FeCl₃) and glass were used as the precursor, solvent, dopant source and substrate respectively. Zinc acetate (Zn(COOCH₃)₂$\cdot$2H₂O) has been widely used in literature as the precursor of ZnO growth^{[3, 10, 12-14]}. According to the certain proportion, Zn(COOCH₃)₂ and FeCl₃ are first dissolved in water at room temperature for Zn$_{1-x}$Fe$_x$O, ($x = 0, 0.01$, 0.03 and 0.05) thin film preparation. The glass substrates are cleaned in acetone and dried properly. Then a clean substrate with a suitable mask ($2.5 \times 2$ cm$^{2}$) is put on the susceptor of the heater.Process parameters like air pressure, deposition rate, distance between tip of the spray nozzle to substrate etc. affect the structural properties of the films. Hence, all these parameters are kept constant throughout deposition of all films. The distance between the tip of the spray nozzle and the surface of the glass substrate is kept at 25 cm. The pressure is maintained at 0.5 bar and temperature of heated substrates is kept at around 350 ℃. During the deposition process, compressed air of carried gas is passed through a tube at the pressure of 0.5 bar which creates a vacuum at the tip of the spray nozzle. This creation of vacuum leads to sucking the solution from the tube after which the spraying of the solution starts onto the heated substrate. Thus, the chemicals in the solution are vaporized, react on the substrate surface after reaching it and Fe doped ZnO thin films are deposited on the glass substrate.

3. Experimental results and discussion

3.1 Microstructure

The morphology of Fe doped thin films were observed by SEM. Figure 1 shows the SEM micrographs of the pure and Fe doped ZnO films. Figure 1(a) shows more or less uniform flake-like nano-sized grain structure of ZnO. The overall morphology of the layers seems to be due to the growth and clustering of initial nuclei. Figure 1(b) shows randomly oriented and distinct flake shape structure. Figure 1(c) shows the film surface increases the random distribution of grain, coarseness and in homogeneity with increasing Fe doping. Figure 1(d) shows small crystallites and agglomeration of the grain particles for Fe doped (5 at%); films exhibit dense packed and homogeneous growth in the entire surface.

3.2 XRD analysis

The structural evolution of the deposited ZnO and the Fe-doped ZnO films has been followed by X-ray diffraction (XRD) measurements using a diffractometer with Cu K$\alpha$ radiation ($\lambda$ = 1.54178 Å). The unit cell of the crystal was found to be hexagonal wurtzite with the presence of dominant peaks at (100), (002), (101), (102), (110), (103) and (112) as shown in Fig. 2. It is in good agreement with the standard JCPDS card no. 36-1451 and Ref. [10, 18]. Furthermore, the intensities of different diffraction peaks are different, which indicates that the growth of ZnO in various planes is different and the growth is anisotropic. From Fig. 2, it is also observed that for pure ZnO there is high intensity of peaks, predominantly (100), (002) and (101) with low full width at half maximum (FWHM). This suggests good crystalline quality of undoped ZnO thin film. With 1 at% Fe-doped ZnO thin film, the intensity of peaks are reduced and FWHM is increased compared to undoped ZnO thin film revealing that the crystalline quality is weakened by introducing doping. However, when Fe-doping concentration is increased to 3 at%, the intensity of peaks are enhanced again and the FWHM is decreased. This suggests the crystalline quality of ZnO thin film is greatly improved by 3 at% Fe-doping. On the other hand, when Fe-doping concentration is above 3 at%, such as 5 at%, crystalline quality is weakened again indicated by reduction of peak intensity and increased FWHM. This result is also supported in the previous studies of influence of Fe-doping concentration on the crystalline quality of ZnO films. In Ref. [3], Fe-doped ZnO thin films were prepared by spray pyrolysis using FeCl₃ as the dopant and XRD results showed that the 3 at% Fe-doped ZnO provided the strongest peak and the smallest FWHM which is similar to our experimental result. However, the results are very different for other studies with other dopant sources and growth methods. In Ref. [11], Fe doped ZnO thin films were prepared by sol-gel method on Si and glass substrates and the best crystalline growth was obtained for 1 at% Fe-doping; higher doping caused weakening of crystalline properties in turn. For Fe-doped ZnO thin films on Si and glass substrates by magnetron sputtering using Fe-chips as the dopant in Ref. [15], XRD results showed that the intensity of diffraction peak decreased and the FWHM increased with increasing Fe-doping concentration, implying the crystalline quality of ZnO thin film declines gradually. The different results of the above experiments occurred due to different distribution of Fe ions in ZnO thin films as well as the difference in preparation techniques and deposition parameters of ZnO thin films. In our experiment, 3 at% Fe-doping improves the crystalline quality of ZnO thin film compared to lower doped ones because the creation of new nucleation centers from the dopant atoms is favorable for the growth of ZnO crystals. However, the crystalline quality is degraded again when Fe-doping concentration is 5 at%. This is probably connected with the following two factors: (1) the newer nucleation centers reach saturation^[3]; (2) due to the difference of ionic radius between Fe$^{x +}$($x = 2$ or 3) and Zn$^{2+}$, when a large number of Fe$^{x +}$ replace Zn$^{2+}$ in lattice sites substitutionally, lattice distortion is intensified, resulting in larger strain in the films and consequently affecting the normal growth of ZnO crystals. However, in our experiment, as FeCl₃ was used as the dopant source, it is likely that the lattice distortion has occurred due to substitutional replacement of Fe$^{3 +}$ in Zn$^{2+}$ lattice sites where there exists significant difference of ionic radius between Fe$^{3 +}$ and Zn$^{2+}$.

3.3 Structural analysis

3.3.1   Influence of Fe-doping on diffraction angle

From Fig. 3, it is observed that for the 0, 1, 3 and 5 at% Fe-doped ZnO thin films, the dominant peaks for different planes like (002) and (101) shift toward bigger angle direction. The shifting of peaks is attributed to the strain in the film. Increasing Fe-doping concentration introduces strain in the films causing a shift in diffraction angle. A similar result is also reported by Kim^[1] and Xu^[3]. However, the results reported by Wang^[15] and Chen^[17] showed that the (002) peak gradually shifts towards smaller angle direction with the increase of Fe-doping concentration. On the other hand, an interesting phenomenon has been reported by Luo^[16]: the peak shifting for (002) occurred towards bigger angle direction when Fe doping was $\leqslant 1.2$ at% and to lower angle direction when Fe doping was $\geqslant 2.6$ at%. The difference of the above results is mainly associated with the valence state of Fe ions in ZnO^{[3, 16]}.

It is known that both the valence state and the ionic radius are the major factors determining the solubility of the dopant. In Zn$_{1-x}$Fe$_x$O alloys, Fe ions need to have a valence of $+2$ in order to properly substitute Zn$^{2 +}$ ionic sites while maintaining charge neutrality. When Fe$^{3 +}$ ions coexist with Fe$^{2 +}$ ions in Zn$_{1-x}$Fe$_x$O, the Fe$^{3 +}$ ions are expected to distort the lattice structure for holding charge neutrality^[1]. However, regarding ionic radii, the ionic radius of Fe$^{2 +}$, Fe$^{3 +}$ and Zn$^{2 +}$ are 0.078, 0.068 and 0.074 nm, respectively. Thus, the ionic radius of Fe$^{2 +}$ is larger than that of Zn$^{2 +}$ by 5.4%, whereas that of Fe$^{3 +}$ is smaller than Zn$^{2 +}$ by 8.1%. Hence, due to the difference of ionic radii between Fe$^{2 +}$ or Fe$^{3 +}$, when they replace lattice sites in place of Zn$^{2 +}$ substitutionally, both will result in different types and magnitudes of strain. In experiments of Wang^[15] and Chen^[17], the XPS analysis showed that Fe ions exist in the form of Fe$^{2 +}$ in the Fe-doped ZnO thin films where both groups prepared the films by magnetron sputtering using Fe pieces as dopant. The group of Kim^[1] also prepared Fe-doped ZnO thin films by magnetron sputtering but they found Fe$^{2 +}$ or Fe$^{3 +}$ coexisting in the films. On the other hand, the group of Xu^[5] found the dominant existence of Fe$^{3 +}$ in the films prepared by sol-gel method using Fe(NO₃)₃$\cdot$9H$_2$O as dopant. Thus, it is observed that depending on the factors like deposition techniques, Fe dopant sources (e.g. Fe(NO₃)₃$\cdot$9H$_2$O, Fe pieces, etc.) as well as the native point defect density in ZnO thin films, the doped Fe in ZnO can exist either in the form of Fe$^{2 +}$ or Fe$^{3 +}$, or even coexist in the form of Fe$^{2 +}$ and Fe$^{3 +}$. If Fe$^{2 +}$ and Fe$^{3 +}$ coexist in ZnO thin films, the molar ratio of Fe$^{2 +}$/Fe$^{3 +}$ will determine the type of strain. When Fe ions exist in ZnO mainly in the form of Fe$^{2 +}$, due to its larger ionic radius than that of Zn$^{2 +}$, it will lead to compression strain in the film^{[3, 15]}. A visualization of this situation in XRD patterns is shifting of the peak towards smaller angle direction. On the contrary, if Fe ions exist in ZnO mainly in the form of Fe$^{3 +}$, due to its smaller ionic radius than that of Zn$^{2 +}$, it will lead to tensile strain in the film, producing a shifting of the peak towards bigger angle direction in XRD patterns. This is more evident by the experiment of Luo^[16] of magnetron co-sputtering on Zn target with Fe pieces, where Fe$^{3 +}$ substituting for Zn was obtained when Fe doping was $\leqslant$ 1.2 at% and existence of Fe$^{2 +}$ when Fe doping was $\geqslant$ 2.6 at% and the peak angles from XRD shifted to higher and lower values accordingly. In our experiment, the Fe-doped ZnO thin films demonstrated the shifting of peaks towards bigger angle direction with the increase of Fe-doping concentration. Based on the above analyses, it is concluded that Fe ions in our prepared ZnO thin films exist mainly in the form of Fe$^{3 +}$. This is consistent with the results of Xu^[3] who also obtained existence of Fe$^{3 +}$ using a Ferric compound dopant source, Fe(NO₃)₃$\cdot$9H₂O, whereas we also used another Ferric compound dopant, FeCl₃.

3.3.2   Influence of Fe-doping on lattice parameters

In Fig. 4, it can be observed that increasing doping caused a reduction of interplanar spacing (d) value of the film. Table 1 also shows that in doped films, all the lattice constants a, b $(a=b)$ and c axis had reduced value compared to pure ZnO film. The ionic radius of Fe$^{3 +}$ is smaller than that of Zn$^{2 +}$, suggesting that when Fe$^{3 +}$ replaces Zn$^{2 +}$ in the lattice substitutionally, it results in a smaller interplanar spacing (d) value and lattice constants than that of undoped ZnO film^{[1, 19]}. Thus, the reduction of the d value and lattice constants are attributable to the existence of the Fe$^{3 +}$ ions with high density in the present films.

3.3.3   Influence of Fe-doping on grain size

The crystallite sizes (D) of the films are estimated using the Scherer formula^[20]:

$\begin{equation} D=\frac{k\lambda}{\beta \cos\theta}, \end{equation}$

(1)

where k is a constant taken to be 0.89, $\lambda$ is the wavelength of X-Ray used (1.54178 Å) and $\beta$ is the FWHM peak of XRD pattern, Bragg angle, 2θ.

The dislocation density ($\delta$), defined as the length of dislocation lines per unit volume, are estimated using the equation^[20]:

$\begin{equation} \delta=\frac{1}{D^2}. \end{equation}$

(2)

Strain ($\varepsilon$) of the thin films is estimated using the equation^[20]:

$\begin{equation} \varepsilon=\frac{\beta \cos\theta}{4}. \end{equation}$

(3)

The effect of strain in the ZnO films introduced by Fe dopants can be further illustrated from grain size (D), strain ($\varepsilon$) and dislocation density ($\delta$) given in Table 1.For the samples with Fe-doping concentration of 1 at%, grains tend to decrease in size because of the large strain in the films that affects the normal growth of ZnO. However, when Fe-doping concentration is 3 at%, grain size increases greatly because of lower strain. This illustrates again that the best crystalline structure is for 3 at% Fe doping as previously explained from XRD analysis. With higher doping 5 at%, the grain size decreases again with greater stress, leading to poor crystallinity as assumed.

It should be mentioned that the strain value ($\varepsilon$) depends on both $\beta$ and $\cos\theta$ as shown in Eq. (3). With increased doping, although Bragg's angle ($2\theta$) shifts to a higher value giving lower $\cos\theta$, the strain is not decreased accordingly as it is also dominated by $\beta$. From XRD analysis, the smallest FWHM and hence, the lowest $\beta$ for Fe doped ZnO films are obtained for 3 at% doping (Table 1). Hence, strain gets lower for 3 at% doping. For higher doped film, however, the effect of lower $\cos\theta$ is surpassed by large $\beta$ leading to high strain.

Dislocation density ($\delta$) also resembles a measure of crystallinity. Among the doped films, 3 at% doping gives the smallest $\delta$ indicating the best crystalline structure compared to higher or lower doped ones as expected.

4. Optical analysis

4.1 Influence of Fe-doping on transmission spectra

Optical characterization has been performed with UV spectroscopy using UV spectrophotometer (Shimadzu UV-3100, Japan) with a photon wavelength range 200-1100 nm. The transmission spectrum of the ZnO thin film grown on glass substrates is shown in Fig. 5. It exhibits that pure ZnO has a good transmittance in the visible range ($\approx 62\%$-83%). However, doping is a major factor affecting the optical property of ZnO thin films^[3]. With 1 at%, crystalline quality is degraded resulting in decrease of transmittance. On the other hand, improved crystalline quality provided better transmittance with 3 at%. This high transmittance is attributed to the good crystal structure which eliminates light scattering. Higher doping of 5 at% leads to poor transmittance again. This can be attributed to increase in photon striking with increase in carrier concentration in a poor crystal structure. Thus the optical study also supported our structural analysis. Decrease of transmittance in the visible range with decreasing crystalline quality at high Fe concentration is also supported by Ref. [17].

5. Conclusion

In this work, structural and optical properties of Fe-doped ZnO thin films prepared by spray pyrolysis have been investigated. Based on the research results of Fe-doped ZnO materials reported by us and other groups, it is considered that the valence state of Fe plays a great role in the variation of properties of ZnO thin films. The XRD analysis result of our experiment suggests that the doped Fe ions exist mainly in the form of Fe$^{3 +}$ as FeCl₃ is used as dopant. SEM and UV spectroscopy reveal that introducing doping primarily deteriorates structural and optical properties although 3 at% Fe-incorporation can improve the crystalline quality and transmission of ZnO film. However, more Fe-incorporation again deteriorates the crystalline quality and visible region transmission. Considering the structural and optical properties, the fabricated Fe-doped ZnO thin films prepared by a cost effective method, i.e spray pyrolysis, can be used as a suitable material for visible region transmission in optoelectronic application.

Acknowledgements: The authors would like to acknowledge the laboratory facilities provided by the Department of Physics, Bangladesh University of Engineering and Technology, for conducting the research reported in the paper. The authors are thankful to Materials Science Division, Bangladesh Atomic Energy Commission Division (AECD) for providing support of XRD analysis and Bangladesh Council of Scientific and Industrial Research (BCSIR) for SEM and UV spectroscopy.