Structural, optical, photoluminescence and magnetic investigation of doped and Co-doped ZnO nanoparticles

We report here the structural, optical, photoluminescence (PL), and magnetic investigation of Zn1−x−yFexMyO nanoparticles. The lattice constants and crystallite size are decreased by Fe, followed by a further decrease up to (Fe + M) = 0.30. A compressive stress is approved and the size of particle is between 180 and 277 nm and follows the sample order of ZnO, (Fe + Cu), (Fe + Ni), and Fe. Although a single value of energy gap (Eg) is found for pure and Fe-doped ZnO, two values of Eg (Egh and Egl) are found for the co-doped samples. The Eg is generally increased by Fe, followed by a further increase for the Cu-series, whereas it is decreased for the Ni-series. The refractive indices nK and nT proposed by different methods are generally decreased by Fe, followed by a further decrease for both series. Although Fe doped ZnO depressed the density of carriers (N/m*), it increased again for the co-doped samples. The residual dielectric constant ϵL is decreased by Fe, followed by an increase for the Cu-series, but it is decreased for the Ni-series. The loss factor tan δ increases slightly with Fe, followed by an increase for the Ni-series, but it decreases in the Cu-series. A significant depression of optical conductivity σopt by Fe was obtained, followed by a further decrease which is higher for the Cu-series. The PL shows four visible emissions. Interestingly, an IR emission at about 825 nm is only obtained for the co-doped samples. Furthermore, the blue emission (Iblue) was higher than UV (IUV), [(Iblue/IUV) > 1], but it is greater for the Ni series than the Cu. Although ZnO exhibits diamagnetic behavior, the Fe and co-doped samples exhibit ferromagnetic with higher magnetization for the Ni-series than the Cu. The current results recommend the co-doped samples in nanoscale for some of advanced devices.


Introduction
The n-type ZnO has been used in optoelectronic, gas sensor, solar cell, and spintronic devices due to its stability at ambient conditions. (Manikandan et al. 2017;Ragupathi et al. 2013;Padmapriya et al. 2016;Ali and Hakeem 2015). Additionally, ZnO can be used as a specific component in magneto, photo, spin-electronics, and microwave devices when Zn is doped by 3-d transition metals (TM) (Mustaqima and Liu 2014;Djerdj et al. 2010;Yang 2013). Unfortunately, further research is still needed for the co-doped ZnO as compared with individual dopants.
The electron-hole (e-h) pairs are generated when the valence band (VB) electrons are excited by absorbing of photons (Othman et al. 2017). ZnO holds towards blue and UV optical devices due to their wide band gap E g (3.2 eV) of ZnO and larger Urbach energy (60-100 meV) (Liu et al. 2010;Tariq et al. 2019). The UV-blue photoluminescence (PL) band emission of ZnO represents the near band edge (NBE) of charge carriers. The NBE emission can be delineated as a photo-excitation of charge carriers from VB to CB, resulting in an exciton (e-h pairs) formation and subsequent radiative recombination of e-h pairs. In addition, ZnO with additives exhibits good PL centres due to their multi-color emission, which is necessary for white light emitting diodes (LEDs) (Poornaprakasha et al. 2020;Xu and Cao 2009). This is why the PL spectrum of ZnO is characterized by two main peaks (a sharp UV peak at 380 nm and broad deep band emission between (400-600 nm) (Shan et al. 2004;Sedky et al. 2022). The broad deep band emission is due to deep centers, which are caused by intrinsic and extrinsic defects. In addition, some of whole visible region of different coloured could be obtained by defects (Wang et al. 2006).
In order to create ferromagnetic (FM) at a high Curie temperature (T C ) close to RT for spintronic devices, ZnO doped with TM have been early used (Mohamed et al. 2023a). RTFM is caused by exchange spin interaction between free delocalized carriers and localized TM (Kittilstved et al. 2005).Therefore, a lot of work has been investigated through the magnetic moment of ZnO: TM as well as the stable polarized state exhibited by TM (Mohamed et al. 2023b;Sedky et al. 2022a). The T c in ZnO: TM is very sensitive to phase purity, doping type and morphologies. However, the T c has decreased when ZnO co-doped with magnetic ions like Co, Mn, Fe, and Ni (Mandal et al. 2006;Zhang et al. 2007). Likewise the RTFM are affected by oxygen vacancies (Ahmed 2017). For clarifying the origin of RTFM, the ZnO co-doped by two different ions in nanoscale will be presented.
The recent advancement of various quantum size effects in nanotechnology implies that the majority of novel devices are based on pure and co-doped ZnO nanomaterial by shrinking their size to the nanoscale (Aljaafari and Sedky 2020;Dutta et al. 2003;Lopez-Quintela et al. 2003). For example, the optical absorption properties of ZnO nanoparticles are found to be more suitable for many optical devices due to their significant UV reflection as compared to ZnO micro-particles. These are beneficial for ZnO applications in skin care products, optical devices, dye-sensitized and solar cells (Jadhav et al. 2013;Amin and Sedky 2019;Zhang et al. 2016;Karmakar et al. 2010;Wojnarowicz et al. 2015). Unfortunately, further research is required for more advantage for the high stability of co-doped ZnO to be satisfied for advanced devices (Wojnarowicz et al. 2016;Neena et al. 2016).
Recently, Sedky et al. presented a comparative study on the structural and optical properties of Zn 0.90-x Fe x M y O bulk ceramics (Abdullahi et al. 2015;Al-Naim et al. 2021a, 2021b. Unlike the articles mentioned, the particle size has been minimized and structural, optical, PL, and RTFM properties are well evaluated. Regardless of the behavior of the obtained parameters against co-doping content, the phase purity of ZnO was improved, the crystallite size was decreased to about 10 nm and ZnO bond length was increased from 1.974 Å to 2.064. Å. The micro-strain was increased two times and the residual stress was changed from tensile to compressive like-behavior. Interestingly, the average size of nanoparticle domains or grain agglomerations was depressed to a few hundred nanometers and did not exceed 400 nm.

Experimental details
Zn 1−x−y Fe x M y O samples with various M, x,y were synthesized by using the solid-state reaction method. The powders of ZnO, Fe 2 O 3 , CuO and NiO are mixed and calcined in air for 16 h at 950 °C. ZnO,Zn0.1Fe,Zn0.1FeCu0.05,Zn0.1FeCu0.10,Zn0.1FeCu0.20,Zn0.1FeNi0.05,Zn0.1FeNi0.10,and Zn0.1FeNi0.20 were sintered at 1000 °C for 16 h before being slowly cooled to RT. The (Fe + Cu) samples are called the Cu series for simplicity, whereas the (Fe + Ni) are called the Ni series. After that, the sizes of the pellets with equal quantities are minimized to the nanometer scale by a mechanical ball milling machine (200 rpm rotation speed and 20-min duration time). The phase purity and particle size are respectively tested by XRD using Cu-Kα radiation and JEM (1230 TEM) with a 100 kV. The optical properties were tested against wavelength (200-1000 nm) using double beam UV-visible-NIR spectrophotometer (Jasco V-570, Japan). The PL measurements are made by Jasco FP-6500 spectrofluorometer of 150-W Xenon arc lamp and 5 nm bandwidths. The emission spectra were measured at an excitation wavelength of 300 nm. Finally, magnetic measurements (M(H) loops) of the samples were taken using vibrating sample magnetometer (VSM Lakeshore 7410) over a field range of up to + 10 KOe.

Structural analysis
The XRD patterns of the samples shown in Fig. 1 indicated that most of peaks belong to ZnO wurtzite structure [ICSD 01-079-0208 (space group P63mc)]. The ZnO and Zn0.1FeNi0.05 samples are free from any extra lines, while an extra line with little intensity is formed for the remaining samples as listed in Table 1. The 2θ-angles of the extra lines are about 53.26°, 39.82°, and 42.88° for Fe, Cu series, and Ni series, respectively. The average intensities I avg of the (100), (002), and (111) intense lines listed in Table 1 are decreased by Fe, followed by an increase as co-doping increases to 0.10 and then decreases to 0.20, but they are higher for the Cu series than the Ni. Lattice constants a and c, and unit cell volume V listed in Table 1 are reduced by Fe, followed by a further reduce as the content of co-doping increases, but they are higher for the Cu series than the Ni (Sedky 2014;Chow et al. 2013;Aksoy et al. 2010;Seetawan et al. 2011;Wojnarowicz et al. 2018;. The size of crystallite D hkl given by; D hkl = K cos (Sedky 2014) and listed in Table 1, is decreased by Fe doping, followed by a further decrease as the content of co-doping increases to 0.10 and then increases to 0.20, but it is higher for the Ni series than the Cu. The micro strain ε m calculated by; m = cos 4 , and listed in Table 1 is increased by Fe, followed by an increase with more co-doping content to 0.10, and then a decrease to 0.20, but it is higher for the Cu series than the Ni as reported Muhammad et al. 2019;Srinivasulu et al. 2017). However, the lattice strain ε L given by; ε L 456 Page 4 of 24  Table 1 for the two series indicate compressive stress.
The micrographs and histograms of the ZnO, Zn0.1Fe, and Zn0.1FeCu0.1and Zn0.1FeNi0.1 samples shown in Fig. 2 contain agglomerations of different shapes and sizes of irregular grains. The sizes of grains are between (150-400 nm), (250-1500 nm), (200-800 nm) and (200-1000 nm). The sizes (D TEM ) against the high counts of particles are 180, 295, 266, and 277 nm, inconsistent with those obtained for D hkl . This means that D hkl and D TEM are respectively controlled by the ionic size and the ability of the nanoparticles agglomerations, which generally follow the order of (Fe + Cu), (Fe + Ni) and Fe.

Optical measurements
Figure 3a-c plots absorbance (A) against wavelengths (λ). It is seen that A was decreased by Fe-doping, and then decreases for the Cu series, while it increased for the Ni series. It is expected that some of the shallow levels are introduced inside the band gap by (Fe + Cu) co-doping, and consequently, the absorption edge exhibits a red shift as Cu content increases. However, the vice is versa for the Ni series. Two optimum regions of A close to or below 400 nm are evident for all samples, owing to a significant increase in A as λ decreases. The second maximum, however, is missing from the A-curves for pure and Fe samples. This is due to increasing free carrier density as the λ decreases towards near-UV ~ 400 nm (exciton peaks) (Ali et al. 2005). As given in Table 2, the A max occurs at λ (1) max = 249, 256 nm for pure and Fe samples and decreases to 253, 251, 238 nm for the Cu-series, but it is increased to 292, 314, 294 nm for the Ni series. The second A max is obtained at λ (2) max = 379, 368, 354 nm for the Cuseries and 414, 436, 410 nm for the Ni series. This means that the required energy for creating the e-h pairs is generally increased by Cu, but it is also increased by Ni.
The absorption coefficient α is determined by, α = 2.303 ln(A/t), where t is the cuvette diameter. Therefore, the E g was obtained for allowed direct transition using Taucs equation (Abdalla et al. 2003;Wang et al. 2003); However, the plot of (αhυ) 2 against the hυ is used to get the E g (see Fig. 4a-c). Two significant absorption edges are only observed for the co-doped samples of both series, indicating two electronic transitions. The values of E g of the first and second transitions are called E gh and E gL since that (E gh > E gL ).
The values of E gh and E gl listed in Table 2 and shown in Fig. 5 are generally increased by Fe, followed by a further increase for the Cu-series up to (Fe + M) = 0.30. While, they increased for the Ni-series at (Fe + M) = 0.15 Ni followed by a decrease up to (Fe + M) = 0.30. The ΔE = (E gh − E gl ) values are respectively 1.60, 1.40, 1.52 eV for the Cu-series and 0.85, 0.90, 0.25, eV for the Ni series, in which ΔE decreases as M increases, but it is higher for the Cu series than the Ni. Anyhow, two E g are also recorded for n-type semiconductors Ismail et al. 2011). However, the decrease in E g for the Ni-series may be due to some defects formed in the host materials, thereby creating insulating states in the forbidden gap inside the band structure. While the increase in E g for the Cu-series may be due to producing high carrier concentration, some of them may be blocked, and increases the E g as obtained, which is required for high power operation (Yu et al. 2001;Sagadevan et al. 2017a).
Additionally, the refractive indices (n T and n K ) are examined in terms of E g as follows (Kumar and Singh 2010;Tripathy 2015); Table 2 Wavelengths (λ 1 , λ 2 ) against maximum absorbance, energy gaps (E gh , E gL ), refractive indices (n K , n T ), carrier density (N/m*) and residual dielectric constant (ϵ L ) for the samples   Table 2 are decrease by Fe, followed by a further decrease for both co-doped series, but the values of n K are higher than those of n T . The lattice dielectric constant represents the n against λ is determined by Abdel-Khalek et al. (1178); Zeyada et al. 1960): where c, e, ε o , m*, N, k and R denote light speed, free space permittivity, charge and effective mass of an electron, free carrier density, extinction coefficient and reflectance, respectively. The (n 2 -k 2 ) against λ 2 plot shown in Fig. 6a give the ϵ L and N/m* as listed in Table 2 and also shown against (Fe + M) content in Fig. 6b, c. The Fe doped ZnO depressed the (N/m*) from 3.69 × 10 55 to 0.62 × 10 55 (g.cm −3 ), but it increased again to 1.23 × 10 55 (g.cm −3 ) for the co-doping, which slightly supports them for the use of optoelectronic. However, a significant increase of (N/m*) for the 0.85Zn0.1Fe0.05Cu sample is not clear at present, but it may related to the clean phase purity of the samples as indicated in the XRD analysis. This behavior indicates that the electronic polarization of ZnO is generally reduced by co-doping for both series, as reported (El-Desoky et al. 2014). ϵ L was decreased by Fe, followed by an increase for the Cu-series, but it is decreased for the Ni-series as well

Fig. 5 Energy gap as a function of (Fe + M) content for the samples
Page 11 of 24 456 as E g . The increase/decrease of ϵ L as M increases for the Cu/Ni-series, can be attributed to oxygen vacancies (Lin et al. 2013).
The optical (σ opt ) and electrical (σ elec ) conductivities can be evaluated by Sedky et al. (2021); The behavior of both σ opt and σ elec with the photon energy hν for the samples is shown in Fig. 7a, b. σ opt is almost the same with increasing hν to 3 eV, but it slightly increases above this energy. In contrast, a significant depression of 0.1 Fe was obtained, followed by a further decrease for both series, but the rate of decrease was higher for the Cu-series than the Ni. The value is reduced from 5.5 × 10 10 s −1 for ZnO to 4 × 10 10 s −1 for a 0.9Zn0.1Fe sample. On the other hand, σ ele has regular behavior, but its values are 10 4 times lower than σ opt . In addition, it was decreased by 0.1Fe, followed by an increase for the Cu-series, but it decreased for the Ni. This trend is due increasing the free carriers by co-doping which support the conductivity.
Furthermore, the dependence of dissipation factor tan δ on the real r ( r = n 2 − k 2 ) and imaginary part i ( i = 2nk ) of dielectric constant, according to the following relation (Mott 1970 As shown in Fig. 8, the tan δ slightly increases by Fe, followed by a further increase for the Ni-series up to M = 0.30, whereas it decreases for the Cu-series. In addition, it is generally decreased when increasing the hν for all samples. However, increasing the quality factor (q = (1/ tan δ) for the Cu-series is required for the use of super-capacitor, while its decrease for the Ni-series is convenient for normal capacitors.

Photoluminescence (PL) measurements
The PL intensity of the samples shown in Fig. 9a, b shows four visible emissions which are required for light emitting diodes. Interestingly, an extra five IR emission peak is obtained only for the co-doped samples. As listed in Table 3, a little bit of shift to higher/lower values of λ as the content of co-doping M increases could be observed, but not exceed 9 nm (λ UV , λ blue , λ green , λ red , λ IR are the wavelengths against the PL emission peaks, respectively). Further, the doping has decreased the PL of ZnO, but it is higher for the Ni series than the Cu, which can be considered evidence of more states of defects induced by the co-doping, due to reducing the crystalline quality of the ZnO with M content. However, there are two very sharp near-UV and blue peaks recorded at 409 and 464.5 nm for ZnO and 410.5 and 466 nm for Zn0.1Fe samples. In addition, they are recorded between (411.5-415 nm), (467.5-469 nm) for the Cu series and between (401-407 nm), (459.5-467 nm) for the Ni series. However, the PL of blue (I blue ) is always higher than that of UV (I UV ), as shown in Fig. 9c, [(I blue /I UV ) > 1], but it is higher for the Ni series than the Cu.
However, the UV emissions (3.088-2.991 eV) are related to the E g of ZnO, whereas the visible are due intrinsic defects of the vacancies and interstitials of both O and Zn (V o , V Zn , O i Zn i ) (Vanheusden et al. 1996;Fan et al. 2005). The violet emissions are controlled  (Tatsumi et al. 2004). The blue emission (2.702-2.647 eV) is due to radiative transition of an electron from Zn i shallow donor to V Zn acceptor (Heo et al. 2005). The highest broad peak of green emission (2.288-2.241 eV) is caused by recombination of donor V o− by acceptor V Zn (Murphy et al. 2006;Sedky and Mohamed 2013).The red emission (2.027-1.988 eV) is attributed to the co-doped ZnO (Sagadevan et al. 2017b;Sedky et al. 2012). The lowest IR emission, (1.516-1.499 eV) is due to excitation of (3d/4f) lowest state, in consistent with ZnO doped by Tm 3+ (Jadwisienczak et al. 2002). However, the obtained emissions are good evidence that the co-doped ions are successfully incorporated into ZnO as reported (El-Said Bakeer 2020; Layek et al. 2016;Van Dijken et al. 2000;Van Dijken et al. 2000;Mohamed et al. 2021). Figure 10a, b depicts the magnetization of the samples against a magnetic field (M-H). ZnO exhibits perfect diamagnetic behavior as reported, whereas clear ferromagnetic is obtained for both series (Ateia et al. 2015). The hysteresis loop for Fe samples (0.9Zn0.1Fe) is weak and, therefore, it can be classified ferromagnetic or super-magnetic. Interestingly, the magnetization was increased with the further addition of M up to 0.20 (co-doped samples), but it is higher for the Ni-series than the Cu. But with an increase of M to 0.30, a decrease in magnetization has occurred for both series. The magnetization of the Ni series decreased, but is still higher than that of M = 0.20 (0.8Zn0.1FeNi0.05). In contrast, it is depressed for the Cu-series and becomes similar to the Fe sample (0.9Zn0.1Fe). The saturated magnetization M s versus M content shown in Fig. 11a and listed in Table 4 was not obtained for ZnO due to diamagnetic behavior. However, it becomes 0.171, 0.238, 0.296, 0.178 (emu/g) for the Cu-series, and 0.171, 0.291, 0.450, 0.343 (emu/g) for the Ni-series. Although M s was decreased for M = 0.30 samples of both series, it was still higher than that of Fe samples, especially for the Ni-series.

RTFM measurements
To be more specific, the M s = [M s (co-doped)-(M s (Fe)] values listed in Table 4 are 0.067, 0.125, and 0.007 (emu/g) for the Cu-series and 0.12, 0.279, and 0.172 (emu/g) for the Ni-series.
Firstly, the decrease of M s for M = 0.30 sample in both series is related to the XRD extra lines as a result of solubility limit discussed above. This behavior indicates that RTFM is controlled by the degree of phase purity obtained through synthesis and co-doping content. Secondly, Ni is ferromagnetic with effective magnetic moment (μ = 2.83μ B ). When Zn +2 are replaced by ‫‬ Ni +2 in either tetrahedral or octahedral coordination, the T c of ZnO are controlled by the interaction of Ni bound magnetic poltroons within the localized spins and free carriers (oxygen vacancies). Nevertheless, it has reported by the authors that M s of 0.20 Ni doped ZnO is about 0.085 (emu/g) (El-Bassuony and Abdelsalam 2017), which is 5-times lower than of (0.8Zn0.1FeNi0.10), and indicating strong mutual participation between localized spin of co-doping of Fe and Ni as compared to individual doping by Fe or Ni even in case of decreasing (N/m*) for Ni-series stated above. Although Cu is perfect diamagnetic, M s of Cu-series are 1.7times more than of Fe, which is very interesting point. This is due increasing N/m* of Cu-series, which may also increase the mutual interaction between Fe localized spin and free carriers. To support our thinking, the M s of Cu-series is about 1.5 lower than of Ni-series, which indicates that RTFM of ZnO is controlled by the introduced free carriers beside the localized spin of TM. To our knowledge, M s of 0.7Zn0.1FeCu0.20 was decreased to 0.178 (emu/g), which is approximately similar to Fe doping, 0.171 (emu/g).  (Srinet et al. 2013;) are listed in Table 4. Furthermore, the M-H straight line slope (dc susceptibility χ dc ) versus M is also shown in Fig. 11b. All of them are increased by the co-doping, but they are higher for the Ni-series than the Cu, which strongly supports the RTFM for all doped and co-doped samples as expected. (Coey and Chambers 2008;Samanta et al. 2018). The observed ferromagnetism at RT seems to be the combined effect of the interaction of TM spin at the tetrahedral sites with charge carriers of oxygen vacancies to form bound magnetic polarons (Coey and Chambers 2008;Samanta et al. 2018;Mohamed et al. 2022;Yong-Sheng et al. 2010).
Recently, the RTFM of ZnO can also be enhanced through a co-doping mechanism (Ram and Negi 2016;Sedky et al. 2022b). For example, an induced RTFM could be obtained as Fe increases in Zn 1-x Co 0.05 Fe x O, but unfortunately it is lower than that of Zn 1-x (Fe + Cu) x O [95][96]. They indicate that Zn 1-xx Cu 0.05 Fe x O does not have a tendency to form the Fe-O-Co ferromagnetic cluster. In contrast, Zn 1-x (Fe + Cu) x O has a tendency to form the Fe-O-Cu ferromagnetic cluster of the double-exchange mechanism with the charge transfer between Fe and Cu. Although this behavior inconsistent with the present work, all of them are correct because they fixed the Co/Cu content at 0.05 and increased the Fe content. Regardless, the local spin of Fe, which increases for both series, the rate of interaction, of Fe induced spin with charge carrier due to Cu-ion, is higher than Fe + Co induced spin with charge carrier. This will eventually support the conflict with our work, which strongly highlights the present investigation.
Let us now compare the present analysis with those reported but in microscale (Chow et al. 2013). Firstly, the phase purity of ZnO was improved and the V was increased from (46.397-47.876 Å 3 ) to (53.479-54.108 Ǎ 3 ). While the average crystallite size, D hkl , was decreased from about 26 nm to about 10 nm. The ε m was increased to twice times and  Fig. 9 a PL intensity as a function of wavelength for the pure and (Fe + Cu) samples b PL intensity as a function of wavelength for the pure and (Fe + Ni) samples. c The blue shift (I blue /I uv ) as a function of (Fe + M) content for the samples the residual stress was transferred from tensile to compressive like-behavior. Interestingly, the D TEM was depressed to a few hundred nanometers not exceed 300 nm, which is about 8-times lower than D SEM . Interestingly, the λ max and the E g were shifted to lower values. They are decreased form (264-340 nm), (2.851-3.974 eV) to (249-314 nm), (3.02-3.88 eV). Moreover, two E g are obtained for the co-doped samples. Similar behavior was obtained for the (N/m*) and ϵ L which they were also decreased from (0.86 × 10 55 -221.40 × 10 55 (g.cm −3 )), (1.96-23.96) to (0.62 × 10 55 -3.96 × 10 55 (g.cm −3 )) and (1.626-12.55). Based on the above, we concluded that the behaviors of materials nanoscale are convenient for most of advanced applications rather than microscale.

Conclusion
Structural, optical, photoluminescence, and magnetic properties of Zn 1−x−y Fe x M y O nanoparticles were investigated. The average nanoparticle size is between 180 and 277 nm and follows the sample order of ZnO, (Fe + Cu), (Fe + Ni) and Fe. Two values of E g are for the co-doped samples. The E g was increased by the co-doping, but it is higher for the Cu-series than for Ni. Similar behavior is obtained for (N/m*), ε L and q-factor. A blue shift [(I blue /I UV ) > 1] is obtained for all samples, but it is higher for the Ni series than the Cu. A noticeable RTFM is clearly obtained for the Fe and co-doped samples with evaluated magnetization parameters. These findings strongly recommend the (Fe + Cu) series in nanoscale for high power operation, solar cell and supercapacitor devices. In contrast, the (Fe + Ni) series is recommended for light-emitting diodes and spintronics.