Md Shafayatul
Islam
a,
Koushik Roy
Chowdhury
a,
Sheikh Manjura
Hoque
b and
Ahmed
Sharif
*a
aDepartment of Materials and Metallurgical Engineering, Bangladesh University of Engineering and Technology, Dhaka 1000, Bangladesh. E-mail: asharif@mme.buet.ac.bd
bMaterials Science Division, Atomic Energy Centre, Dhaka, Bangladesh
First published on 6th February 2024
The potential of oxide-based dilute magnetic semiconductors (DMSs) for use in spintronics and magneto-optic devices has garnered a lot of attention over the years. However, the optical and magnetic behavior of these DMSs is challenging to navigate due to the complicated interactions of intrinsic defects. In these contexts, the current research takes a comprehensive look at the pristine and simultaneously multiple d-block cation (Cr, Fe, Ni, Co, and Zn)-doped CuO nanocrystals (NCs) to explain the defect interactions inside the lattice. Structural analysis revealed a highly crystalline monoclinic crystal structure in the C2/c space group. The phase stability of CuO NCs was found to decrease with increasing dosages of dopants, ultimately forming a secondary phase of Cu metal. Diffuse reflectance spectroscopy (DRS) spectra showed narrowing of the optical band gap, attributing it to the presence of impurity states between the conduction band minimum (CBM) and valence band maximum (VBM) as a result of doping. These impurity states can inhibit carrier recombination. Both pristine and doped CuO NCs showed ferromagnetism at ambient temperature with a paramagnetic tail at higher fields. This paramagnetic tail, explained based on the thermomagnetization curves, corresponds to the easy flip of magnetic cations below room temperature. All the ferromagnetic features of CuO NCs may be traced back to the exchange interaction between the spins of magnetic ions, mediated by carrier-trapped vacancy centers. Interestingly, doped oxides showed improved ferromagnetism when used at moderate concentrations, which is related to the creation of a larger number of bound magnetic polarons (BMPs). However, higher dosages of dopants lower the concentration of BMPs by delocalizing the carriers from the defect centers.
In addition, CuO is a potential candidate for oxide-based dilute magnetic semiconductors (DMSs), which are promising future materials for spintronic applications due to their versatility in spin transport, spin detection, and spin filtering.11 For these applications, possession of room-temperature ferromagnetism (RTFM) is essential for DMSs, which is arguably intriguing and the most mysterious problem in magnetism. Hence, the discovery of RTFM in CuO has been a core area of research interest, which is probably affected by defect equilibrium in the crystal lattice. A feasible mechanism for inducing RTFM could be the modulation of defects by incorporating dopant elements such as d-block cations. These dopants can tailor ferromagnetism as well as the position of Fermi energy levels through the interaction of the 3d orbital electrons, thus tuning the magnetic, optical, and electric properties of the host CuO.
Several researchers have reported that doping CuO with d-block elements causes a dramatic shift in its physicochemical properties and forms new types of materials with enhanced physical and chemical characteristics without any major modification of the host CuO structure that has interesting spin-dependent electrical, optical, and magnetic properties. These new properties are attributed to modifications of the bandgap, alteration of the electronic structure, sp–d exchange interactions that involve the d-sub-levels of d-block ions, and holes in the valence band of the host semiconductor. For example, band gap values were modified and found to increase as the doping concentration of Cr increased in CuO.12 In addition, Cr can induce RTFM in CuO.13 Further, it was observed that Co-doped CuO nanoparticles show superparamagnetism at room temperature.14 One important finding is that Fe dopants in CuO prefer to align anti-parallel to one another, while Fe dopants in larger numbers tend to align parallel to one another.15 The absorption peak of Ni-doped CuO nanoparticles moves towards a higher wavelength, i.e., a redshift, and the magnetization goes up with Ni concentration up to 5 mol% doping, and further increases in Ni content decrease the magnetization.16 With increasing the doping concentration of Zn, the band gap edge shifted towards the longer wavelength region.17 However, the origin of the improved ferromagnetic behavior of CuO doped with non-magnetic cations (such as Zn2+) has been the subject of relatively few investigations. These reports of RTFM in undoped, magnetically doped, and non-magnetically doped CuO all point towards the significance of defects in generating ferromagnetic ordering as well as customizing bandgaps.
In general, a significant amount of oxygen vacancies, along with other point defects, are common on the surface of nanoscale semiconductors like CuO because of the size effects.18 The point defects strongly affect the energy states, which can modify the optical and magnetic characteristics. Therefore, point defects like lattice oxygen as well as cation vacancies, interstitials, and free charge carriers act as controlling factors for tailoring the bandgap and transport properties of dilute magnetic semiconductors (DMSs) like CuO. The next natural step going forward would be to study how the incorporation of these defects in different concentrations and configurations affects the optical and magnetic properties of this structure.
Moreover, it is observed that the physical properties of CuO NCs are also affected by the morphology, structure, and size of the nanocrystals, which in turn depend on the specific synthesis method.19,20 Several research groups synthesized CuO NCs by chemical, physical, and biological methods such as hydrothermal, co-precipitation as well as green-mediated combustion, microwave irradiation, and sol–gel.9,21 These methods are capable of producing nanoscale particles with various shapes, sizes, morphologies, and dimensions, such as zero-dimensional (0D) nanoparticles, one-dimensional (1D) nanorods, two-dimensional (2D) complex interconnecting nanosheets, and three-dimensional (3D) nanoflowers, to explore the intriguing application of CuO NCs by tuning different parameters like pH, time, temperature, and capping agents. Among these, sol–gel is of great interest as it is economical, eco-friendly, and has an impact on the size of the NCs and the formation of defects like point defects, dislocations, vacancies, and others.
After reviewing the available literature, it has become clear that little work has been done so far to determine the collective effect of simultaneously doping multiple d-block elements on defect modulation of metal oxides such as CuO. Therefore, pristine and multiple d-block element-doped CuO NCs were successfully synthesized via a modified sol–gel method with varying concentrations in the present work. To determine the experimental origin of ferromagnetism in DMSs and to pave the path for the growth of spin electronic devices, the synthesized samples (pristine and doped CuO NCs) will be connected to physical and structural changes caused by doping. A clear picture of mysterious defect interactions inside the inorganic CuO structure via localization and delocalization of carriers at various defect centers due to defect modulation is depicted. This fundamental study provides useful insights into the broad implications across multiple domains like water splitting, photocatalysis, and dilute magnetic semiconductors. Furthermore, it paves a way to regulate the same monoclinic CuO structure for a variety of applications by tailoring bandgap and room-temperature ferromagnetism through multiple d-block cation doping and also gives a base to understand the bound magnetic polaron interactions in the presence of multiple magnetic and nonmagnetic impurities. Consequently, this study aimed to evaluate the physicochemical properties of pristine CuO and the effectiveness of simultaneously incorporating multiple d-block elements (Cr, Fe, Co, Ni, and Zn) to assess its modified structural, morphological, optical, and magnetic properties extensively via defect chemistry for assigning suitable applications in terms of structural distortion, particle size, bandgap, band-position, and magnetism.
(1) |
(2) |
(3) |
Cu2+ + 2e′′ = Cu | (4) |
(5) |
The necessary explanation of the symbols is included in the ESI.† The merging of the split peak of pristine CuO into a single peak (as shown in Fig. 2b) in doped CuO NCs towards larger angles indicates a slight distortion in the symmetry of the monoclinic system due to the creation of a charge imbalance by Fe3+ and Cr3+ ion doping as a replacement for Cu2+ in the copper oxide lattice.27 Hence, from these defect reaction analyses (eqn (1)–(5)), defects generated in the system can be attributed to the charge imbalance that occurs from Cr3+ and Fe3+ incorporation, which leads to cation vacancies along with oxygen vacancies.
All the diffraction peaks are very intense and sharp, which indicates the strong, well-ordered polycrystalline nature of the synthesized materials.28 However, the intensities of the (11−1) and (111) planes are much higher than those of the rest of the peaks, indicating a prominent orientation of the formed crystallites along either of these directions, which will later be confirmed by TEM observation. For higher doping concentrations of CuO-b and CuO-c, a sharp fall in intensity is noticed, which may happen due to the disorder caused by the disparity in ionic size between the dopant and parent ions. The decrease in peak intensity is attributable to a reduction in the crystallinity of the CuO NCs as doping induces a lattice strain phenomenon in the CuO matrix.29 The peak broadening of the doped NCs compared to pristine CuO is also strain-induced and can be attributed to various reasons, e.g., the formation of smaller particle sizes as well as defects in the lattice structure.29 The calculated values of a, b, and c lattice parameters obtained from the Rietveld refinement, along with the constituent phases, space group, unit cell volume, bond length, bond angle, and reliability (R) factors of the Rietveld refinement, are listed in Table S1 (ESI†). The following equation for monoclinic lattices was used in Rietveld refinement to find the lattice parameters:30
(6) |
As discussed earlier, the amount of dopant present in CuO NCs is inversely proportional to the degree of crystallization. In addition, we have used the Rietveld refined XRD spectra to quantitatively measure the crystallinity of the samples using the following equation:31
(7) |
(8) |
(9) |
Samples | d (Å) (11−1) plane | Crystallite size (Scherrer's formula), D (nm) | Microstrain, ε ×10−3 | Dislocation density, δ × 10−3 (nm−2) | Crystallinity (%) |
---|---|---|---|---|---|
Pristine CuO | 2.525 | 32.94 | 1.8 | 0.92 | 92.18 |
CuO-a | 2.527 | 28.01 | 2.1 | 1.27 | 90.92 |
CuO-b | 2.512 | 24.3 | 2.4 | 1.69 | 87.03 |
CuO-c | 2.514 | 25.1 | 2.3 | 1.587 | 89.31 |
Further, the microstrain was calculated using the following equation:33
(10) |
All these parameters (percentage crystallinity, d spacing, crystallite size, dislocation density, and lattice strain) were calculated corresponding to the characteristic (11−1) peak and are tabulated in Table 1.
Notably, from Table 1, the “d” values corresponding to the (11−1) plane of pristine CuO, CuO-a, CuO-b, and CuO-c are found to be 2.525, 2.527, 2.512, and 2.514 nm, respectively. This variation is due to the slight shift of the (11−1) peak compared to pristine CuO owing to the lattice distortion of the synthesized doped NCs. Higher d spacing between (11−1) planes in pristine CuO and CuO-a along the [11−1] direction ensures fast ion transport between (11−1) planes.34
Moreover, as can be observed from Table 1, the crystallite size first falls when the doping concentration is raised. Dopant ions at the grain boundaries may be to blame for this, since the dopant ions increase the energy barrier for the diffusion of Cu2+ ions and so restrict the expansion of the crystal, creating smaller crystallite size.35 Thus, the role of impurities, i.e., dopant atoms, is to impede grain growth by segregating to the grain boundaries, which might have happened for the CuO-a and CuO-b NCs. However, a further increase in doping concentration (CuO-c) enhances the interaction between the dopant ions, which in turn reduces the effectiveness with which they may settle at the grain boundaries. This leads to enhanced growth of CuO-c crystallites as well as a reduction of microstrain (ε), as observed from the XRD pattern analysis. The reduction in micro-strain is obtained from 2.4 × 10−3 for CuO-b to 2.3 × 10−3 for CuO-c, which signifies a decrease in lattice imperfections, an increase in lattice parameter “a” (as shown in Table S1, ESI†), and the construction of a more dislocation-free lattice, which is confirmed by the reduced value of dislocation density for CuO-c NCs. Therefore, with a larger D, the internal strain in the crystal is at a minimum, and there are fewer dislocations per unit area (Fig. 4).
Fig. 4 Variation of crystallite size, microstrain and dislocation density as a function of doping concentration. |
Using Rietveld refinement data, VESTA was used to calculate the bond lengths and bond angles of the NCs (as depicted in Fig. S1 (ESI†) for pristine CuO). From Table S1 (ESI†), it is seen that changes in bond length show a similar tendency to the lattice parameter “a”. The Cu–O–Cu bond angle shows a decreasing tendency up to the CuO-b doping level, then increases, which is also in good accordance with the evolution of crystal size. This finding suggests that the variation of the lattice parameter “a” and the size of the particles is controlled by the bond angles and bond lengths.
The formation of Cu vacancies to ensure charge balance will decrease the number of Cu–O bonds and lead to a reduction in the Cu–O–Cu bond angle compared to pristine CuO, as evidenced from Table S1 (ESI†). The reduced Cu–O–Cu bond angle and increased Cu–O bond length indicate distortion of the lattice. However, the Cu–O bond length is shown to suddenly drop with greater doping concentrations of CuO-c due to the increased likelihood of dopants residing at the interstitial site. These interstitial dopant ions may influence the lattice via their interactions with other copper or oxygen ions, resulting in a fluctuation in CuO bond length. Hence, dopant concentration is crucial in regulating the intrinsic lattice orientation and defect density of the host system.
Fig. 5 shows no unintentional impurity, except carbon, in both samples. Surface contamination accounts for the occurrence of the C 1s peak in the survey spectra. The high-energy resolution spectrum of the C 1s envelope for pristine CuO as well as CuO-c was fitted with three peaks, as shown in Fig. 6a and b. For pristine CuO, the C–C bond peak was located at 284.8 eV, the C–O–C peak at 286.1 eV, and the O–CO peak at 288.1 eV.36 The binding energy of the C 1s peak at 284.8 eV was utilized as a charge reference in the XPS experiments. For the doped CuO-c sample, regarding these peaks, a small shift in binding energy is observed. Peaks at 284.6 eV, 286.1 eV, and 288.5 eV for CuO-c correspond to C–C, C–O–C, and O–CO bonds, respectively. The shift of the binding energy peaks of the doped sample compared to the pristine CuO can be attributed to the difference in the electronegativity values of Cu and dopant elements.
High-resolution scans of the Cu 2p spectra are shown in Fig. 6c and d. As depicted in Fig. 6c, the pristine CuO NCs consisted of a doublet of Cu 2p3/2 at 933.0 eV and Cu 2p1/2 at 953.0 eV with an energy difference of 20 eV, which is consistent with the standard spectrum of CuO.37 That is to say, the Cu ions undergo oxidation during the reaction. The energy difference between 2p3/2 and 2p1/2 was due to the spin–orbit coupling. In this spectrum, the peaks centered at 941 and 943.7 eV are assigned to satellite peaks for 2p3/2, with 8 and 10.7 eV higher binding energies than those of the main peak. Since Cu2O and Cu do not include shake-up satellites, their existence in Cu 2p spectra may be attributed to CuO NCs.38 This, together with the broad nature of the Cu 2p3/2 peak, confirms the prevailing presence of the Cu2+ state, indicating that +2 is the only valence for Cu ions in undoped NCs. The XPS spectrum of doped Cu 2p is deconvoluted using 20% Gaussian and 80% Lorentzian functions (as depicted in Fig. 8d) into two separated fitted pairs of peaks at 932.3 and 952.2 eV, revealing the co-existence of Cu0 states with Cu2+ at the surface of the CuO-c NCs. Thus, in CuO-c, the Cu 2p spectrum consists of cubic Cu and monoclinic Cu2+ ions, respectively, which was also observed from XRD analysis. However, the peaks of Cu 2p3/2 and Cu 2p1/2 in doped CuO-c NCs are slightly shifted to lower (redshift) binding energies as compared with those of pristine CuO NCs at 932.9 eV and 952.9 eV, respectively. The slight shift of binding energy indicates that the modification in the bond length of CuO, as confirmed by XRD, is due to the incorporation of dopant ions in the CuO lattice.
The high-resolution scan of the XPS spectrum of Cr 2p is shown in Fig. 6e, which indicates the peak of Cr 2p3/2 detected at 576.1 eV for the CuO-c NCs. The peak is near the peak of the Cr3+ state but is different from 574.3 eV for Cr metal, 579 eV for Cr4+, and 576.0 eV for the Cr2+ state.39 This indicates that Cr3+ ions are incorporated into the CuO lattice as dopants. The Cr 2p1/2 peak appears at 585.9 eV. The satellite feature of the Cr 2p3/2 peak overlaps the Cr 2P1/2 component in Cr2O3.
Fig. 6f shows that the Fe 2p spectrum is split into Fe 2p3/2 at 711.8 eV and Fe 2p1/2 at 724 eV due to spin–orbit interaction. Due to interference from a prominent Auger peak of Cu L3M23M23 from copper, it is challenging to further utilize the Fe 2p doublet for quantitative purposes.30 The absence of a peak at 706–707 eV rules out the presence of metallic Fe in doped samples.40 The peaks are in good agreement with those for Fe2O3 (B.E. ∼710–711 eV). These results show that the valence of Fe ions is mostly +3. In other words, Fe3+, rather than Fe2+, takes the position of Cu2+ in the crystal lattice. It is clear why the concentration of cation vacancies and the concentration of Fe ions or Cr ions follow a similar trend. Hence, the more the Fe3+ and Cr3+ ions are replacing Cu2+, the greater the vacancy percentage is needed to accommodate the charge conservation rule.
The signature of cobalt in the NCs was evident from the high-resolution XPS spectrum in Fig. 6g, corresponding to the core levels of two spin–orbit doublets, Co 2p3/2 at 780.3 eV and Co 2p1/2 at 796.18 eV.41 The spin–orbit splitting of 15.9 eV is in good agreement with the literature report of Co, present in the +2 oxidation state. In addition to this, two satellite peaks also occur at higher binding energies (786.1 eV and 802.6 eV).41
In Fig. 6h, the Ni 2p spectrum exhibits two peaks at 855 and 872.5 eV, which are in good agreement with the binding energies of Ni 2p3/2 and Ni 2p1/2, respectively,42 indicating the presence of the Ni2+ state. Besides, the satellite peaks of Ni 2p located at 861.2 and 880.1 eV also prove the existence of the Ni2+ (Ni–O bonds) state.43
The peaks that are visible in Fig. 6i were assigned to Zn 2p3/2 and Zn 2p1/2, which were positioned at 1021.4 eV and 1044.4 eV, respectively. The Zn 2p core electron level is unreliable for distinguishing Zn from ZnO according to XPS literature. For this purpose, the Zn Auger peak (L3M45M45) was recorded as depicted in Fig. S2 (ESI†) and α, the Zn Auger parameter, was calculated (α = BE (Zn 2p3/2) + KE (Zn Auger), where KE is the kinetic energy of the Auger electron). Here, α = 1021.4 + 989.3 eV = 2010.7 eV and fits in with ZnO, of which α values are reported to be in the 2009.5–2011 eV range. Thus, the Auger parameter obtained in this work is much lower than that of metallic zinc (2013.4–2014.4 eV),30 which agrees with reported values for ZnO in the 2009.5–2011 eV range. Therefore, the resulting Auger parameter is much less than the value for metallic zinc (2013.4–2014.4 eV), implying that Zn oxidized (as Zn2+) in the CuO NCs.
Fig. 7a depicts the O 1s spectrum of the pristine CuO sample. The O 1s peak revealed an asymmetrical trend with two broad humps, pointing to the existence of oxygen species with multiple components found in the near-surface area. The high-resolution XPS spectrum of O 1s was fitted using three Gaussian components corresponding to three peaks at 529.5, 531.2, and 533.3 eV. These peaks were assigned to the O−2 ions in the monoclinic structure of CuO (OL), oxygen-deficient regions or defects on the surface created by the doping in the CuO lattice (OV), and the hydroxyl group (O–H), which referred to adsorbed water at the surface.44
Fig. 7 High-resolution spectra of (a) O 1s from pristine CuO and (b) O 1s from CuO-c, along with deconvolution. |
It has also been noted that doping CuO causes a little change in the binding energy of the O 1s spectrum (Fig. 7b). The peak intensities also vary somewhat from one another, most likely because of differences in the number of defects present at various doping levels. The O 1s spectrum of doped CuO-c shows a redshift at 529.4, 531, and 533.3 eV, respectively. The amounts of oxygen vacancies may be inferred from the intensity of the OV peak. We have compared the XPS spectra of the O 1s of CuO-c and pristine CuO to demonstrate the effect of doping on the concentration of oxygen vacancies. The oxygen-deficient area peak of the pristine CuO sample is much weaker than the peak of the CuO-c sample, as seen clearly from the figure. This finding is suggestive of an increase in oxygen vacancies in CuO as a consequence of effective doping element incorporation. Moreover, the relative concentration of oxygen vacancies (OV/O 1s) can be roughly semi-quantitative by its area ratio in the O 1s XPS curves. Consequently, the concentration of oxygen vacancies rises from 31.86% for pristine CuO to 35.33% for doped CuO-c according to the comparison of the high-resolution O 1s XPS spectra. These results indicate that doping may increase oxygen vacancies, which is beneficial for the development of ferromagnetic coupling and the enhancement of magnetism in CuO NCs.8
Therefore, a combination of research results, obtained via XRD and XPS methods, allows us to conclude that the obtained NCs are composed of copper oxide, and for doped NCs, chromium, iron, cobalt, nickel, and zinc atoms are likely to be included in the monoclinic crystal structure of copper oxide.
The SEM images (as shown in Fig. 8a–d) support the formation of irregular polyhedron-shaped fused particles for these samples. This peculiar growth in the grains may be attributed to the uncontrolled aggregation of atoms on the lower energy lattice sites.45 Moreover, during the annealing of the samples, the formation of larger grains by the coalescence of smaller grains might occur, which would increase the size of the particles. It is also apparent from the SEM images that multi-granular structures with strong interconnections are porous. Such porous surface morphology can deliver a greater surface area, making it feasible for gas-sensing device applications and supercapacitor properties. Moreover, the lengths of the grains are found to be inhomogeneous, with a wide range of size distribution.
Doping has caused a significant alteration in the morphology of the prepared CuO particles. In some regions, the big particles were surrounded by smaller nanoparticles for the doped samples. Due to the agglomeration of these tiny particles, lump-like structures were formed. The lumping of powders may be due to moisture produced inside the samples.46 The particles are agglomerated from a few nm to a few μm. This agglomeration minimizes their surface free energy, which signifies the thermodynamically stable growth mechanism, resulting in an even more porous structure, as shown in Fig. 8c. Also, the surface appears to be rough. Distortion and discontinuity of the lattice may be caused by the different radii of the dopant ions and Cu2+, which can then lead to the suppression of grain growth. Therefore, when the concentration of the dopant increased up to CuO-b, the particle size dropped, which is in good accordance with the calculated crystallite size from XRD analysis (as shown in Table 1). However, another reason for this size reduction behavior could be the Zener pinning effect.47 According to this effect, vacancies and other flaws (whose presence is confirmed by XRD and XPS) in the CuO crystals exert an inhibiting force that limits their growth.48
These reductions in particle size with enhanced porosity provide a high volume-specific surface area for better ion intercalation, more electroactive sites, and enhancement of electrolyte access that gives desired electrochemical properties with improved efficiency in attaining higher capacitance. Hence, it is beneficial to employ these doped CuO NCs as the electrode material in order to gain a remarkable cycle ability and superior specific capacitance for supercapacitors.49 In addition, the increased surface area of the doped particles may also have a major impact on the optical and magnetic properties of the samples.50 However, a few particles displayed broken edges as a consequence of incomplete growth during synthesis. On further increasing the doping concentration, the particle size increases, which is analogous to the increase in crystallite size obtained from XRD analysis in the case of highly doped CuO-c NCs.
For all samples, average sizes were determined using a Gaussian fitted histogram in OriginPro software, which are depicted in Fig. 8e–h. In order to better understand the data, log-normal functions were fitted to the histograms. The fitted functions provided an average particle size of 268, 222, 145, and 201 nm for pristine CuO, CuO-a, CuO-b, and CuO-c samples, respectively. Since the size and surface area of sol–gel-derived CuO are crucial to achieving many of the desired physical characteristics in a wide range of applications (sensing, catalysis, etc.), the introduction of dopants into the synthesis process is very encouraging.51
The HR-TEM images of the synthesized NCs shown in Fig. 9 show the presence of ellipsoid-shaped nanocrystals, which are very thin and partially transparent. The HR-TEM images in Fig. 9a–d confirmed that each particle is the accumulation of ultra-small nanoclusters of single crystals after nanocrystal deposition and growth. Hence, very few single crystals are completely separated. This confirms that the actual sizes of these synthesized NCs are very small compared to those measured from SEM images. These nanocrystals also have a tendency to form amorphous boundaries at the junction of the clusters. This phenomenon may be linked to the variation in structural disorder and microstrain induced in the host CuO matrix.
Although the nanocrystals were highly agglomerated and overlapping of crystal fringes occurred due to this high agglomeration, the lattice fringes of some polycrystalline CuO NCs were visible at higher magnification, which are shown in Fig. 9e–h. Fig. 9e presents the calculated interplanar distances of pristine CuO, where the lattice spacing of 0.2516 nm belongs to the (11−1) diffraction planes of monoclinic CuO, indicating preferential growth of these nanocrystals along the [11−1] crystal axis. The calculated values of the doped samples corresponding to the (11−1) plane are 0.2517 nm for CuO-a, 0.251 nm for CuO-b, and 0.2512 nm for CuO-c samples. It should be noted that these observations are in considerable accordance with the XRD results (as tabulated in Table 1). The slightly large interlayer spacing of CuO and CuO-a provides easy diffusion of the electrolyte.52
The crystallinity of the prepared CuO NCs that was previously discussed in the XRD analysis can also be confirmed by the selected area electron diffraction (SAED) pattern of the TEM analysis (as shown in Fig. 9i–l). The polycrystalline nature of the CuO NCs is confirmed by their appearance as concentric rings made up of bright spots with varying orientations. The ImageJ tool was used to determine the diameter of the rings, and the resulting d-spacing values were analyzed. The interplanar distances of pristine CuO NCs, found at d1 = 0.278 nm, d2 = 0.253 nm, d3 = 0.232 nm, and d4 = 0.1866 nm, can be indexed to crystal planes (110), (11−1), (111), and (−202), respectively. This indexing of the SAED pattern corresponding to (110), (11−1), (111), and (−202) planes is consistent with the XRD analysis and the matched JCPDS card number. Thus, the SAED pattern confirms the successful synthesis of tenorite C2/c symmetric CuO NCs. Moreover, for doped CuO-b and CuO-c, an extra ring at an interplanar distance of d = 0.208 and d = 0.21 nm, respectively, indicates the presence of the Cu (111) plane. Misorientation of the fundamental crystallographic domains is indicated by the elongation of several diffraction spots in the diffraction patterns.53 Thus, it can be concluded that doping plays a critical role in altering the morphology of pristine CuO NCs.
The EDX analysis was performed to study the types of elements present in the samples. The elemental analysis results (Fig. S3, ESI†) indicate the presence of Cu and O in pristine CuO. The spectra showed no impurity peaks other than the C peak (due to surface contamination during TEM analysis), which further confirms that the as-prepared nanocrystals are freed from impurities that can emerge from the initial precursors, like nitrogen. For doped samples, the EDX spectral analysis confirmed the presence of Cr, Fe, Co, Ni, Cu, Zn, and O atoms. Hence, dopant incorporation into the structure is confirmed.
The optical properties of pristine CuO and doped CuO NCs were analyzed using diffuse reflectance spectroscopy (DRS) to find the application boundaries of synthesized samples in various fields like photodetectors, photocatalysis, photovoltaics, pollutant removal systems, and energy storage devices.9 The DRS spectra were recorded in the wavelength range of 400–1000 nm at room temperature to predict the energy band structure. Fig. 10a shows the relative amount of reflected light from the samples using reflectance spectra. The reflectance spectra show only around 10% reflectivity in the visible and ultra-violet regions, which justifies the black appearance of the synthesized NCs in sunlight.54 In addition, the observed high infrared reflection can be attributed to the plasmon resonance of NCs.55
Fig. 10 (a) Reflectance vs. wavelength spectra and (b) indirect band gaps of pristine and doped CuO NCs. |
To study the absorption characteristics of nanocrystals, the reflectance spectra were converted into the absorbance mode using the Kubelka–Munk function (F(R)) using the following equation:56
(11) |
F(R)hυ = A(hυ − Eg)n | (12) |
Samples | Absorption edge (nm) | Bandgap (eV) | Refractive index (n) | Dielectric constant (ε∞) |
---|---|---|---|---|
Pristine CuO | 984.69 | 1.26 | 3.09 | 9.55 |
CuO-a | 1008.71 | 1.23 | 3.10 | 9.61 |
CuO-b | 1042.61 | 1.19 | 3.13 | 9.80 |
CuO-c | 1069.58 | 1.16 | 3.15 | 9.92 |
The calculated indirect band gap energy (the electron momentum is not conserved in the transition) for pristine CuO is 1.26 eV, which is notably lower than the previously stated values for undoped CuO.3 The electronic transition from the hybridized O-2p valence band to the Cu-3d conduction band provides a measure of the band gap in undoped CuO. However, in this study, the lower bandgap values of pristine CuO can be attributed to the presence of surface oxygen vacancy states, as observed in XPS. The surface oxygen vacancies can narrow the bandgap for metal oxides and downshift both the VB and CB, indicating higher wavelength adsorption leading to better utilization of the solar spectrum and thermodynamically favorable conditions towards water oxidation (as depicted in Fig. 11) due to the decrease in overpotential associated with this reaction.57,58 Hence, this material is suitable for energy harvesting applications such as sunlight-driven oxygen generation from water.59,60 The photocatalytic potential of pristine CuO NCs was investigated by determining their valence band maximum (VBM) and conduction band minimum (CBM) with the use of the Mulliken electronegativity approach.61 According to this formula,
(13) |
EVBM = ECBM + Eg | (14) |
Fig. 11 Schematic diagram of the calculated band edge position with respect to the H+/H2 potential and O2/H2O potential. |
The bandgap values of doped CuO are 1.23, 1.19, and 1.16 eV for CuO-a, CuO-b, and CuO-c NCs, respectively. Thus, it is found that the optical bandgap is gradually decreased with doping concentration. The bandgap tailoring effect, caused by the existence of copper vacancy defects along with oxygen vacancies (as shown in defect reaction equations in eqn (1) and (2), may be responsible for this decrease in the energy band gap from 1.26 to 1.16 eV. Furthermore, dopants are expected to have a low degree of hybridization for a stable electronic configuration, which in turn may lead to the creation of a unique energy level in between Cu 3d and O 2p, resulting in a narrowed band gap.64 Therefore, the absorption edge is redshifted as a result of the electrical transition from the valence band to the dopant level and subsequently from the dopant level to the conduction band. Another factor that may contribute to bandgap narrowing is the presence of a localized t2g state of the doping element, which can be found in the center of the band gap (when Cr or Fe is used as the dopant) or at the top of the valence band (when Co is used as the dopant).65 Hence, it can be deduced that the band gap values in the doped CuO NCs dropped because the electrical transition between the CB and VB was disrupted by the creation of a number of defect energy bands in the presence of dopants. The absorption edge of pristine and doped CuO NCs was estimated using the equation:66, where h is the Planck constant and c is the speed of light. Strong absorption in the visible region and a continuous redshift with an increase in the concentration of dopants are confirmed by the absorption edge values. The absorption edge along with the band gap of the pristine and doped CuO nanocrystals are listed in Table 2. The increasing redshift also indicates that surface defects are likely to blame for the intra-gap state effects.67 Moreover, the higher visible absorption achieved for the doped NCs leads to higher application efficiency. Entrapment of carriers (e−/h+) occurs because dopant impurities act as trap centers. Carrier entrapment is a phenomenon that may reduce the recombination rate even when the carrier density and quantum efficiency are both increased. This suggests doped CuO NCs might be used in solar cell technology.68 Also, doped NCs have improved degradation performance (Fig. 12) because the narrowed bandgap allows for more absorption of infrared and visible light at a lower recombination rate. When exposed to sunlight, electrons in the valence band transit to the conduction band, creating electron/hole (e−–h+) pairs. By oxidizing water molecules close to the surface, the hole creates hydroxyl radicals (˙OH), and an electron collides with oxygen molecules to create superoxide radicals (˙O2−). The organic contaminants in the water were oxidized and broken down by these free radicals. The response mechanism is as follows:
˙O2− + ˙OH + Organic contaminants → Degradation products | (15) |
Photogenerated electrons are trapped in the impurity band, lowering the rate of electron–hole recombination, enhancing the mean lifetime of electrons.69 As the photo-generated charge carriers migrate to the surface of the nanostructured particles and the oxidation process takes place there, the particle surface area also plays an important role in photocatalysis.70,71 Photocatalysis is enhanced when the particle size of NCs is reduced (due to dopants), since this allows more pollutant molecules to approach the catalyst surface. Moreover, the existence of abundant surface valence states and defects for doped NCs (as shown in Fig. 6) can improve the intrinsic activity of the doped CuO samples.72 However, heavy doping could have been able to fine-tune the optical characteristics further by forming recombination centers due to the delocalization of the carriers, as confirmed by the formation of the secondary Cu phase mentioned in XRD and XPS analyses.
Important factors for integrated optoelectronic devices are the refractive index and dielectric constant of the synthesized nanocrystals. The degree to which a semiconductor transmits incident spectrum radiation is indicated by its refractive index. The Herve and Vandamme relation was used to determine the refractive index (n) of the synthesized materials from the bandgap values as follows:
(16) |
To study the magnetic properties, field-dependent magnetization (M) versus magnetic field (H) curves of the synthesized CuO NCs are displayed in Fig. 13a. The M–H curves obtained at 300 K show narrow magnetic hysteresis loops, indicating a typical weak ferromagnetic nature at lower fields where magnetization does not reach the saturation value even up to the maximum applied magnetic field. This non-saturation and linear behavior in the samples indicate the presence of paramagnetism. If the susceptibility of the paramagnetic (PM) component is known, the contributions of paramagnetism and ferromagnetism can be distinguished. In line with earlier literature, we selected a simple method for calculating the susceptibility of the PM component by calculating the tangent of the top curve (the demagnetizing curve) of the M–H loop at the highest applied field.68 When the PM contribution is subtracted from the total response, all the samples represent S-like curves, showing room-temperature ferromagnetic characteristics. FM and PM components are shown separately in Fig. 13b and c. It is possible that the matrix of NCs is responsible for this paramagnetic component seen in the room temperature M–H curves of CuO samples.76 As depicted in Fig. 13c, PM interaction is highest for the CuO-b sample and lowest for the pristine CuO sample, with similar contributions for the CuO-a and CuO-c samples. This difference in PM interaction with doping concentration can be attributed to the different magnetic transition temperatures for random flips of the magnetic moments present in the matrix, which is shown in Fig. S4 (ESI†) as the relationship between magnetization (M) and temperature (T). With a transition temperature of just 40.1 K, CuO-b initiates paramagnetism at a substantially lower temperature than other NCs, which can be attributed to the greater PM contribution reported at RT for CuO-b.77 This low transition temperature of CuO-b NCs can be attributed to the smallest crystallite size of 21.8 nm compared to other NCs.78 Moreover, the low transition temperature of CuO-b enhances its potential to be used in magnetic hyperthermia.79 The identical contribution of paramagnetism may be explained by the comparable transition temperature of 110.02 K for CuO-a and CuO-c. Hence, the M–T curves confirm that the linear portion of the M–H curves is coming from the presence of paramagnetism in the matrix at RT. As a result, changes in the composition and grain size might affect the coupling between magnetic moments, leading to a modification of the transition temperature.
Fig. 13 (a) M–H hysteresis curves of CuO NCs measured at room temperature, (b) ferromagnetic contribution of NCs and (c) paramagnetic contribution of NCs. |
The saturation magnetization (Ms), remanent magnetization (Mr), squareness ratio (S = Mr/Ms), and coercive field (Hc) are calculated using the hysteresis loops and tabulated in Table 3. The coercivity value of pristine CuO is 69.7 Oe. Upon enhancing the doping concentration from pristine CuO to CuO-b, the coercivity has been on the rise, from 69.7 Oe to 189.08 Oe. Coercivity drops to 55.5 Oe if the doping concentration is increased further (for CuO-c). Coercivity is an indicator of the strength of a magnetic field necessary to overcome the energy of magnetic anisotropy. Since the current samples have little magnetic anisotropy, as noticed by their tiny hysteresis loop, their ferromagnetism is rather pliable.80 The smaller crystallite size of CuO-b usually results in relatively large exchange coupling, J, which is taking place between the core and the surface of the NCs and can be attributed to this larger coercivity.81 High coercivity plays a vital role in biomedical applications.82
Samples | Coercivity, Hc (Oe) | Saturation magnetization, Ms (emu gm−1) | Remanent magnetization, Mr (emu gm−1) | Squareness (Mr/Ms) | M o (emu gm−1) | Susceptibility, χm × 10−6 (emu gm−1 Oe) | m eff × 10−17 (emu) | N × 1015 (cm3) |
---|---|---|---|---|---|---|---|---|
Pristine CuO | 69.7 | 0.1154 | 0.004 | 0.034 | 0.1212 | 2.2 | 13.9 | 0.87 |
CuO-a | 107.7 | 0.2081 | 0.009 | 0.043 | 0.2056 | 7.41 | 9.87 | 2.08 |
CuO-b | 189.08 | 0.1946 | 0.012 | 0.062 | 0.1914 | 9.73 | 6.99 | 2.74 |
CuO-c | 55.5 | 0.0412 | 0.002 | 0.049 | 0.0415 | 6.77 | 5.36 | 0.77 |
Saturation magnetizations were measured to be 0.1154 emu gm−1 for pristine CuO, 0.2081 emu gm−1 for CuO-a NCs, 0.1946 emu gm−1 for CuO-b NCs, and 0.0412 emu gm−1 for CuO-c NCs. In addition, pristine CuO NCs had a remanent magnetization of 0.004 emu gm−1; CuO-a NCs had a remanent magnetization of 0.009 emu gm−1; CuO-b NCs had a remanent magnetization of 0.012 emu gm−1; and CuO-c NCs had a remanent magnetization of 0.002 emu gm−1. These changes in remanent magnetization with doping concentration are involved in domain wall motion.83 Due to this low retentivity, all the samples are defined as magnetically clean materials. The decrease in retentivity to a lower value even than the pristine CuO NCs for CuO-c suggests that CuO NCs can be used for the data storage application. As a measure of its ferromagnetic strength, pristine CuO has a squareness value (squareness of the hysteresis loop) of 0.034. There seems to be a gradual change from a soft magnet to a hard magnet and back to a soft magnet, as the squareness values initially increased to 0.043 and 0.062 for CuO-a and CuO-b, respectively, before decreasing to 0.049 for CuO-c NCs. Hence, doping in CuO can be used as a regulator for the magnetic nature of CuO NCs (Fig. 13a).
The Ruderman–Kittel–Kasuya–Yoshida (RKKY) and bound magnetic polaron (BMP) models are two of the most popular theories used to explain the RTFM.84 Free conduction band electrons are involved in RKKY interactions, and CuO cannot transform from a semiconductor into a metal at such a low doping concentration. This means that DMSs, such as resistive CuO systems, are not a good match for the RKKY model.85 However, XPS measurements and defect reactions demonstrate the presence of point defects such as oxygen vacancies and host lattice cation vacancies in the present samples, suggesting that these point defects can efficiently mediate intrinsic ferromagnetism. The BMP model provides a satisfactory explanation for the ferromagnetic process generated by these defects. Bound magnetic polarons (BMPs), which are considered as an isolated ferromagnetic identity, create orbits with a finite diameter that are similar to those of a hydrogenic Bohr orbit due to the localization of the electrons in the defect centers.86 Thus, depending on the cation interaction with vacancies, magnetism could occur in the CuO NCs via the cation–defect center–cation configuration. By breaking chemical bonds due to heat evaporation, oxygen atoms may escape from lattice locations during the annealing of sol–gel-derived CuO NCs, creating oxygen vacancies. Further, due to air annealing, some vacancies will be filled. This two-contrasting phenomenon leads to an optimum oxygen vacancy concentration. In pristine CuO, electrons released by oxygen vacancies may be “trapped” by Cu2+ ions, leading to the formation of an ionized vacancy (Ov+) state. The s–d exchange interaction among the spin of the 1s1 electron of the Ov+ state and the 3d9 electron of the Cu2+ ions favors long-range FM ordering. At room temperature, a long-range ferromagnetic order is created when BMPs overlap, aligning Cu2+ spins (as shown in Fig. 15).
Fig. 15 Schematic representation of the evolution of the oxygen defect-mediated BMP model with increasing dosage of the dopant. |
The influence of doping magnetic (Cr3+, Fe3+, Co2+, and Ni2+) and non-magnetic (Zn2+) cations in CuO NCs on the evolution of FM characteristics with varying doping concentrations is also analyzed in the present study. These additional magnetic and non-magnetic impurities can enhance oxygen vacancies during annealing, as confirmed by the XPS result of CuO-c. Also, the exchange interaction between Cu2+ and magnetic cations at the surface in the presence of a carrier in the Ov+ state presents a high ferromagnetic coupling energy, even at larger distances. According to some researchers, cation vacancies were also found to contribute to the intrinsic exchange interactions in the DMS system through BMP production, in addition to the oxygen vacancies.87 Therefore, RTFM in doped CuO may also be generated due to the exchange interaction between dopants mediated by Cu vacancies (formed due to the substitution by Fe3+ and Cr3+). In addition to the Cu 3d orbitals and the oxygen vacancies, the O 2p orbitals that originate from the Cu site vacancy may also be supportive of magnetic ordering.88
Moreover, BMPs can also be used to explain the features of M–H curves. The unsaturation nature indicates that BMP concentrations are below the percolation threshold that is required for bulk ferromagnetism; as for the inner region of the NCs, BMPs cannot be formed due to a lack of oxygen vacancies. Each BMP has a significant net spontaneous magnetic moment that aligns rapidly in a magnetic field, leading to a rapid increase in M even at weak H fields, as the d spins in the matrix outside the BMP align only slowly with the H field. Paramagnetically interacting d-spins in the matrix at RT result in a considerably lower moment per d-spin than the BMP. Hence, only at higher fields, the magnetic contribution from the paramagnetically aligned Cu2+ of the core is dominant, which ultimately leads to a paramagnetic tail in the samples.
The BMP model was also used to perform a quantitative study of the magnetization curves by isolating the contribution of the BMPs from that of the PM matrix, determining the true spontaneous moment ms of a single BMP, and calculating the number of BMPs per cm3 (N). In the BMP model, ferromagnetic nature may be satisfactorily explained using the Langevin profile:
(17) |
M = MFM + MPM | (18) |
M = M0L(x) + χmH | (19) |
As noticed from Table 3, the number of BMPs is changing since we are varying the number of donors and magnetic defects. An increased FM in CuO-a and CuO-b indicates the formation of more BMPs through an increase in cation and oxygen vacancies. Electrons trapped in OV+ induce spin orientation for the adjacent dopants, forming a ferromagnetic coupling of the magnetic cation–OV+–magnetic cation. Thus, doping may increase the number of BMPs by increasing the number of defect centers. However, the suppression of BMPs in CuO-c, even in the presence of magnetic impurities and vacancies (as confirmed by XPS), results in a reduction of FM behavior. This suppression of BMPs can be attributed to the delocalization of carriers from the bound polarons due to the increasing mobility of charge carriers at higher carrier concentrations. Thus, it breaks the fundamental structure of polarons. In the present study, XRD and XPS analyses ensure the presence of a secondary Cu phase in CuO-c, which confirms the delocalization of bound electrons between metallic Cu and oxygen vacancies at this higher doping, which were previously localized or trapped. Thus, the defect reaction equation (eqn (4)) can be rewritten more accurately as
Cu2+ + 2e′′ (delocalized) → Cu (metal) | (20) |
Another phenomenon observed from the fitting data is the decrease in effective magnetic moment per BMP (meff) with increasing concentrations of dopants, which in turn decreases saturation magnetization (Ms). This variation in Msversus doping concentration can be attributed to the overlapping of BMPs, as due to overlapping each magnetic moment is shared between two or more polarons, dividing its magnetization effect among the overlapped polarons. Moreover, in doped NCs, the extra dopants can be located in the O2− lattice of the tenorite structure, forming an antiferromagnetic coupling of the magnetic ion–O2−–magnetic ion, which may drastically cut down on the ’magnetic moments of polarons. Also, substitution of Cu2+ ions by non-magnetic Zn2+ reduces the number of exchange interactions inside the polarons as Zn2+ is diamagnetic in nature. Hence, the combination of these factors reduces the effective magnetic moment per BMP (meff) to a greater extent, which becomes more prominent with increasing doping concentration. This also explains the decrease in saturation magnetization for CuO-b (as noticed in Fig. 13b). Despite the presence of a higher number of polarons, the saturation magnetism is less than that in the CuO-a sample because of this reduction in meff, as saturation magnetization is the product of the number of polarons per cm3 and the spontaneous magnetic moment per polaron. Together, our results provide evidence that the dopant plays a significant role in the magnetic moments of CuO NCs and reinforce the relevance of the BMP model in understanding the underlying mechanism responsible for the ferromagnetism of dilute magnetic semiconductors.
Based on the defect reactions, XPS analysis, and values generated from fitting the BMP model, a schematic representation of the BMP model is created, as shown in Fig. 15. In addition to the magnetic polarons, the CuO matrix also contains some randomly distributed isolated ions. It has been shown that oxygen vacancies (Ov) are the most dominant defects in CuO. Neighboring BMPs overlap and interact with one another via the Ov+ site when the concentration of localized oxygen vacancies rises.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ma00987d |
This journal is © The Royal Society of Chemistry 2024 |