Bi_0.9Ho_0.1FeO₃/TiO₂ Composite Thin Films: Synthesis and Study of Optical, Electrical and Magnetic Properties

Abstract

A visible light active Bi_0.9Ho_0.1FeO₃ nanoparticles/TiO₂ composite thin films with different mol.% of Bi_0.9Ho_0.1FeO₃ were successfully prepared via non-aqueous sol-gel method. The incorporation of 5, 10 and 20 mol.% Bi_0.9Ho_0.1FeO₃ nanoparticles in the precursor solution of TiO₂ brings modifications in the functional properties of the composite thin films. XPS analysis indicates that interdiffusion of Fe³⁺, Ho³⁺, Bi³⁺/Ti⁴⁺ ions through the interfaces between Bi_0.9Ho_0.1FeO₃ nanoparticles and TiO₂ matrix reduces the concentration of Ti³⁺ ions. X-ray diffraction analysis affirms that TiO₂ and Bi_0.9Ho_0.1FeO₃ retain anatase and orthorhombic phase respectively in composite films. The composite thin film containing 20 mol.% Bi_0.9Ho_0.1FeO₃ nanoparticles exhibits the most prominent absorption phenomenon in visible region and has significantly reduced indirect band gap of 2.46 eV compared to that of pure TiO₂ (3.4 eV). Hall effect measurements confirm that the resistivity of composite film increases by ∼2.33 orders of magnitude and its carrier concentration decreases by 1.8 orders of magnitude at 5 mol.% Bi_0.9Ho_0.1FeO₃ nanoparticles addition compared to those of pure TiO₂ film. Moreover, the pure film exhibits diamagnetism, whereas the composite films have both large ferromagnetic and small diamagnetic components. The findings in this research justify that the composite film can be a potential candidate for making improved photocatalyst, resistors and spintronic devices.

Introduction

TiO₂ as powder or thin film has been comprehensively investigated due to their excellent photo-chemical stability, low cost and non –toxicity¹. However, pure TiO₂ responds only to ultraviolet (UV) light which comprises only 4% of the total sunlight. The absorption occurs in the UV light due to large band gap (3.2 eV) which eventually restricts its practical application in degradation of organic pollutants such as dyes, detergents and pesticides etc^1,2,3,4. Because of diamagnetism and low resistivity, TiO₂ has become a concern in the field of spintronics, resistors, sensors and multistate memory devices^5,6. These limitations have galvanized endeavors to enhance photocatalytic activity, induce magnetism and raise resistivity of TiO₂.

One of the attempts is to substitute a portion of the titanium with different electropositive atoms. Transition metals (e.g. Fe, Mn and Cr etc) are the most commonly used constituents that induce visible light absorption^2,3,4. Besides, transition metal (TM) doped TiO₂ has exhibited improved ferromagnetism at room temperature^7,8. Moreover, Mn and Fe dopants in TiO₂ have increased the resistivity significantly^7,8. Another endeavor involves substitution of oxygen with electronegative atoms (N, C and S) which aids TiO₂ to absorb visible light efficiently⁹. However, the introduction of substitutional elements creates scattering centers which decrease the photochemical activity. It has been reported that substitutional TM serves as a center for charge carrier recombination². Besides, it has been controversial as to whether Co or Fe doped TiO₂ is a true magnetic semiconductor or if the effect is due to the clustering of a ferromagnetic second phase^10,11. Indeed, according to these reports doped TiO₂ has failed to exhibit magnetism to any greater extent.

Since, the physical quantities like electrical, magnetic, and optical properties come in the category of sum or product properties, many have tried to make composites of TiO₂ to utilize this criteria^12,13. Magnetic properties have been improved by the fabrication of TiO₂/Fe₃O₄/SiO₂ or Fe_xO_y-TiO₂ composites^14,15. Similarly the nanostructured ZnO/TiO₂ has been found to improve the optical properties, even though, ZnO is a wide band gap semiconductor¹⁶. Nowadays, the visible light activity of TiO₂ is increased by coupling it with other narrow bandgap semiconductors. The proposition is that coupled semiconductors form a hetero-junction structure which can transfer electrons from an excited small band gap semiconductor to the other attached one in the case of proper band potentials. In₂O₃¹⁷, BiFeO₃¹⁸ or CdSe co-sensitized TiO₂¹⁹ heterostructures have been reported for showing better visible light activity. Multiferroic BiFeO₃ (BFO) deserves a special mention for its narrow band gap (2.2 eV) and chemical stability which have made it a suitable visible- light responsive photocatalyst¹. The coupling of TiO₂ with BFO has been accomplished through different synthesis processes. For instance, core-shell structured BFO/TiO₂ by hydrolysis precipitation approach¹, deposition of TiO₂ on ferroelectric BFO substrate by pulsed laser deposition²⁰, growth of TiO₂ nanofibers on BFO nanoparticles by electrospinning¹⁸. These reports focus on enhancing visible light activity of TiO₂. Subsequently, several studies ponder on improving electrochemical energy storage capacity and solar energy conversion efficiency of titania by anchoring BFO nanoparticles on TiO₂^21,22.

The magnetic and electrical properties (e.g. resistivity, carrier concentration and mobility) still remain unreported with BFO nanoparticles due to its various limitations. Literature survey indicates that BFO nanoparticles are antiferromagnetic²³ and have higher leakage current due to formation of oxygen vacancies (OVs)²⁴. Recently, doping with tri-positive rare-earth ions (R³⁺) at the Bi site has been proposed to mitigate some of the issues mentioned above²⁵. Previous reports suggest that 10 mol.% Holmium (Ho³⁺) doping at Bi site increases the magnetization by 6 times and exhibits higher resistivity compared to pure BFO²⁵. Since Ho³⁺ is a rare earth metal it is expected that Bi_0.9Ho_0.1FeO₃ (BHFO) nanoparticles will have better absorbance in the visible region and lower band gap than BFO²³. The aforementioned advantages make BHFO nano particles a good candidate for incorporation in TiO₂.

TiO₂ thin film is better candidate for spintronics, sensors and magnetic memory devices creating faster, smaller and more energy efficient devices compared to titania nanoparticles, nanowire, nanorod or nanofiber. Both undoped and doped TiO₂ thin films have been of immense interest due to their enhanced photocatalytic activity, better electrical properties and room temperature ferromagnetism^3,7,26,27. To the best of our knowledge the optical, magnetic and electrical properties based on incorporation of BHFO nanoparticles in TiO₂ thin film have not been explored yet. Herein motivated by the above concerns, BHFO nanoparticle/TiO₂ composite thin films were prepared by non-aqueous sol-gel method and their structural, magnetic, optical and electrical properties were investigated in detail. The current study revealed that composite thin films could be a more efficient visible light absorber than pure TiO₂. Furthermore room temperature magnetism and improved electrical properties of this composite thin films have been found.

Experimental Section

Materials

Bismuth nitrate pentahydrate (Bi(NO₃)₃.5H₂O, Merck -India), iron nitrate nonahydrate (Fe(NO₃)₃.9H₂O, Merck-India), holmium nitrate pentahydrate (Ho(NO₃)_3.5H₂O, Sigma Adrich-USA), citric acid (C₆H₈O₇, Merck -India) and ethylene glycol (C₂H₆O₂, Merck -India) were used for the synthesis of BHFO nanoparticles. In this case, C₆H₈O₇ acted as the chelating agent to complex the metal cations and C₂H₆O₂ was added as polymerization agent²³.

The solutions of pure TiO₂ thin film and BHFO nanoparticles/TiO₂ composite thin films were prepared using titanium(IV) n-butoxide (Ti (O-nBu)₄, Aldrich-USA) as precursor and n-butyl alcohol (n-BuOH, Merck-India) as solvent. Besides, acetylacetone (C₅H₈O₂, Loba chemie) was added in the solution as chelating agent to decrease the reactivity of Ti(O-nBu)₄²⁸. Glacial acetic acid (CH₃COOH, Qualikems) was used further to begin hydrolysis via an esterification reaction.

Synthesis of BHFO nanoparticles

BHFO nanoparticles were synthesized using a modified sol-gel method described elsewhere²³. Shortly, for a typical 1 g BHFO powder synthesis process stoichiometric proportion of Bi(NO₃)₃.5H₂O (0.003 mol), Fe(NO₃)₃.9H₂O (0.0033 mol), Ho(NO₃)₃.5H₂O (0.00033 mol), C₆H₈O₇ (0.0067 mol) and C₂H₆O₂ (10 ml) were dissolved in 400 ml deionized water. Consecutively the solution was heated under continuous stirring at 75–85 °C for 4 h to obtain gel. The gel was dried at 100 °C for 24 h in a drier to obtain precursor xerogel. The ground precursor xerogel powders were annealed at 500 °C for two hours with a heating rate of 3 °C/min to obtain BHFO nanoparticles. BHFO pellets (thickness 0.11 cm and diameter = 1.3 cm) were prepared by mixing precursor xerogel powders with PVA binder followed by pressing (5 tons pressure) and annealing at 500 °C with a heating rate of 3 °C/min to measure electrical properties.

Preparation of pure TiO₂ thin films and BHFO nanoparticles/TiO₂ composite thin films

The detail synthesis process of pure TiO₂ thin film is delineated elsewhere²⁸. Briefly, first n-BuOH (0.0884 mol) and C₅H₈O₂ (0.0015 mol) were mixed and, then Ti(O-nBu)₄ (0.005 mol) was added to the solution. Subsequently this mixture was stirred for 30 min at room temperature. CH₃COOH (0.001 mol) was slowly added into the alkoxide solution and stirred continuously for another 30 min. Concentration of the final solution was 0.5 M and its color was yellowish.

Thin films containing 5 mol.% BHFO (marked as T⁹⁵B⁵), 10 mol.% BHFO (marked as T⁹⁰B¹⁰) and 20 mol.% BHFO (marked as T⁸⁰B²⁰) were prepared using non aqueous sol-gel method. To synthesize composite films, molar amount of n-BuOH, C₅H₈O₂, Ti(O-nBu)₄ and CH₃COOH was kept similar as that of pure thin film. Firstly 0.0442 mol of n-BuOH was taken and calculated amount of BHFO nanoparticles was mixed into it for every composite film. The whole mixture was dispersed vigorously in an ultrasonic bath for 30 min to avoid the aggregation of BHFO nanoparticles. Then C₅H₈O₂, Ti(O-nBu)₄ and rest of n-BuOH (0.0442 mol) were added into this mixture and it was stirred for 30 min on a stirring plate at room temperature. This mixture was then taken into ultrasonic bath in which CH₃COOH was added dropwise. The final mixture was under intensive stirring for 2 h in an ultrasonic bath. The prepared mixture eventually turned into brownish color.

The brownish and yellowish solutions were spin-coated onto the 2.5 × 2 cm² glass substrate at 2000 rpm for 30 s and dried at 200 °C for 10 mins. During coating brownish solution was in stirring state continuously so that BHFO nanoparticles did not settle down. Micro pipette was used to ensure that specific amount of solution was taken every time. The spin-coating and drying processes were repeated three times. The pure and composite films were annealed in static air at 500 °C for 2 h with a heating rate of 3 °C/min.

Characterization

XPS measurements were carried out on the PHI Quantera II spectrometer and peak fitting was performed by Origin Pro. XRD (PANalytical Empyrean X-Ray Diffractometer system) was used for phase analysis of nanoparticles and thin films utilizing a Cu x-ray source (wavelength: Kα₁ = 1.540598 Å and Kα₂ = 1.544426 Å). Rietveld refinement was done using FULLPROF software²⁹. Field emission scanning electron microscope (FESEM: JEOL, JSM, 7600F) was employed to monitor the morphologies of nanoparticles and thin films and to determine the thickness of these thin films. Energy dispersive X-ray spectroscopy (EDS) was used for mapping the elements present in the thin films. The optical properties were evaluated using UV/Vis/NIR spectrometric measurements (Perkin Elmer, Lambda 1050). To measure the magnetic property of the thin films, 5 mm × 5 mm size sample was cut from every thin film substrate. Electrical characterizations of nanoparticles were carried out by coating both sides of the pellet with silver paste and its conductivity was measured using Precision Materials Analyzer (Radiant Technologies, Inc.: P-PMF, PMF0215-377). Hall measurement was carried out using the van der Pauw configuration (HMS-3300). The four points contacts were made by indium tin alloy soldering which enabled us to obtain the carrier concentration, resistivity and mobility of thin films. The room temperature magnetic property was determined by vibrating sample magnetometer (VSM: EV-9 Microsense).

Results and Discussion

The chemical composition and electronic structure of the as synthesized BHFO, TiO₂ and T⁸⁰B²⁰ was determined by performing XPS analysis. Figure 1(a) depicts the XPS survey spectrum which clearly shows the presence of all the main constituent elements like Ti, Bi and O. Here, all the spectra were corrected for the C 1s peak appeared at 284.6 eV. However, Fe 2p peaks near 720 eV were not easily detectable by XPS nonetheless the EDX analysis (see Fig. S4) suggests the abundance of Fe in composite thin films. The low sensitivity of Fe to XPS analysis may lead to its absence in the survey spectra. However, the high resolution XPS scanning of BHFO nanoparticle depicts the peak corresponding to Fe having low intensity compared to other elements see Fig. S1. The high resolution elementary XPS peaks of Ti 2p and O 1s core levels observed for TiO₂, T⁸⁰B²⁰ and BHFO after curve- fitting are presented in Fig. 1(b–f). The binding energies of the Ti 2p_3/2 and Ti 2p_1/2 are found to be nearly 458.4 and 464.1 eV respectively, giving a spin orbit splitting of ~5.7 eV between the two core levels, which corresponds to the 4+ oxidation state of Ti in anatase phase^30,31. Moreover, the appearance of extra peaks at around 457.8 eV and 463.3 eV confirms the presence of Ti^3+ 31. The peak position of Ti 2p in T⁸⁰B²⁰ has shifted negligibly. It is important to note here that the peak areas ratio for Ti³⁺ to Ti⁴⁺ given by (Ti³⁺ 2p_3/2 + Ti³⁺ 2p_1/2) _Area/(Ti 2p_3/2 + Ti 2p_1/2)_Area in TiO₂ and T⁸⁰B²⁰ are 0.348 and 0.231 respectively. The diminishment of the area ratio in T⁸⁰B²⁰ implies that the concentration of Ti³⁺ has decreased. The interdiffusion of A³⁺ (here A³⁺ represents Bi³⁺, Fe³⁺ and/or Ho³⁺) from BHFO to TiO₂ across the interface could lead to a reduction in the concentration of Ti³⁺ ion. The substitution of Ti⁴⁺ in TiO₂ by A³⁺ generates holes to maintain charge neutrality which oxidizes an existing Ti³⁺ to Ti ⁴⁺ according to the following reactions decreasing the density of electron hopping sites.

$${A}^{3+}\mathop{\to }\limits^{Ti{O}_{2}}{A}_{Ti}^{^{\prime} }+{h}^{\bullet }$$

(1)

$$T{i}^{3+}+{h}^{\bullet }=T{i}^{4+}$$

(2)

Figure 1

(a) XPS survey spectra of BHFO nanoparticles, TiO₂ film and T⁸⁰B²⁰ composite thin film. (b,c) Ti 2p spectra of TiO₂ and T⁸⁰B²⁰. (d–f) O 1s spectra of BHFO, TiO₂ and T⁸⁰B²⁰.

Combining Equations (1,2)

$${A}^{3+}+T{i}^{3+}\mathop{\to }\limits^{Ti{O}_{2}}{A}_{Ti}^{^{\prime} }+T{i}^{4+}$$

(3)

Figure 1(d) displays the XPS spectra of O 1s for BHFO, which can be deconvoluted into two peaks at around 529.4 eV and 531 eV. The lower binding energy peak is affiliated with the intrinsic O 1s core spectra and the higher energy peak is attributed to the oxygen vacant sites in BHFO. The XPS spectrum of O 1s core level for TiO₂, as shown in Fig. 1(e), can also be de-convoluted into two symmetric Gaussian peaks centered at 529.6 and 532.2 eV²². The low binding energy peak at 529.6 eV of O 1s can be assigned to the −2 oxidation state of oxygen³¹. The enhancement of peak intensity at around 532 eV can be ascribed to the presence of both the oxygen vacancy related defects (OV) and the hydroxyl group (${{\rm{OH}}}^{-}$) together in T⁸⁰B^20 22,31. The appearance of ${{\rm{OH}}}^{-}$ in TiO₂ is normal since it is potential photocatalyst^2,4. It is noteworthy here that the ratio of the two peak areas (OV/${{\rm{OH}}}^{-}$)_Area/(O 1s)_Area in TiO₂ and T⁸⁰B²⁰ are 1.83 and 2.20 respectively. The increment in area ratio could be due to an increment in ${{\rm{OH}}}^{-}$ group concentration in T⁸⁰B²⁰. This indicates that the composite film could be a better candidate for making improved photocatalyst than many other ceramic thin films.

Figure 2 shows room temperature XRD patterns of nanoparticles, pure thin film and composite thin films. The obtained pattern of TiO₂ as depicted in Fig. 2(a) shows satisfactory correspondence with anatase phase. The diffraction peaks at 2θ = 25.2° (101), 37.8° (004), 48.0° (200), 53.9° (105) and 55.1° (211)) are matched well with the standard diffraction data of tetragonal crystal structure of TiO₂³². Thus, it is confirmed that the annealed TiO₂ film deposited on glass substrates has anatase phase with no trace of other polymorphs of TiO₂ (e.g., rutile and brookite).

Figure 2

XRD patterns of the prepared pure film, composite films and nanoparticles.

On the other hand, XRD pattern of BHFO as shown in Fig. 2(e) is in line with previous report of Ho³⁺ doped BFO ceramics³³. In addition to the desired spectra one extra peak [marked by asterisk (*)] becomes visible which indicates the presence of impurity phase. To uncover the crystal structure and the impurity phase, Rietveld refinement of BHFO was carried out using the FULLPROF software²⁹. The refinement corroborates that BHFO crystallizes in a orthorhombically distorted perovskite structure having Pnma space group and the extra peak in the XRD pattern belongs to Bi₂Fe₄O₉ impurity phase. However, refinement indicates that Ho substituted BFO has different bond angles (Fe-O2-Fe = 151.33°, Fe-O1-Fe = 147.91°) and bond lengths (Fe-O1 = 1.49 Å, Fe-O2 = 1.82 Å) compared to that of undoped BFO²³. The alternation of these parameters is expected to have profound influence on the magnetic and optical properties of BHFO. Weight fraction of the phases, lattice parameters and Rietveld agreement factors listed in Table S1 and fitted plot of XRD pattern as Fig. S2 are provided as Supplementary Information.

Looking at the XRD patterns for composite thin films (see Fig. 2(b–d)) it can be immediately appreciated that intensity of the peaks of BHFO increases with an increase mol.% of BHFO nanoparticles. In particular, no impurity peak can be discerned in any of the composite films. This observation solidifies that the anatase phase of TiO₂ is retained in the films and addition of nanoparticles does not introduce any phase change. Another feature of the composite thin films is to be observed- the diffraction peaks of different planes of composite films have shifted and the lattice parameters have changed with respect to pure one (see Table 1). The positions of diffraction peaks (101) and (200) were determined using Pseudo -Voight function in high score plus software and lattice parameters were calculated using the following formulae³⁴

$$\frac{1}{{d}_{hkl}^{2}}=[{h}^{2}+{k}^{2}+{l}^{2}(\frac{{a}^{2}}{{c}^{2}})]\frac{1}{{a}^{2}}$$

(4)

where d_hkl spacing has been calculated using the Bragg’s law.

Table 1 Position (2θ) and lattice parameters values obtained from XRD for pure film and composite films.

The peak shifting of composite films may be attributed to the diffusion of Fe³⁺, Ho³⁺ or Bi³⁺ ions into TiO₂ across the interfaces of BHFO and TiO₂. Table 1 shows that diffraction peaks (101) and (200) have shifted to higher angle for T⁹⁵B⁵ and this film has the lowest lattice parameters. According to a previous study, this phenomenon manifests the introduction of tensile stresses due to the interstitial incorporation of doping ions in the crystal lattice of TiO₂³⁵. Some other previous studies also report that Fe ions can go either in octahedral interstitial sites or in the substitutional positions of TiO₂ lattice^36,37. Since the ionic radii of Fe³⁺ (0.785 Å) is lower than that of Bi³⁺ (1.14 Å) and Ho³⁺ (1.015 Å)³⁸, Fe³⁺ ions have the highest chance of interstitial incorporation in TiO₂. On the other hand, the diffraction peaks of T⁹⁰B¹⁰ and T⁸⁰B²⁰ has shifted to the lower angle and they have larger lattice parameters compared to the pure film (see Table 1). These can be ascribed to the generation of compressive stresses as a result of substitutional incorporation of Bi³⁺ (1.14 Å), Ho³⁺ (1.015 Å) or Fe³⁺ (0.785 Å)^35,38. The XPS analysis also supports this phenomenon. The comparatively larger ionic radii of Bi, Ho and Fe compared to Ti⁴⁺ (0.745 Å) ion³⁸ possibly cause an expansion of the crystal lattice and concordant shift in the TiO₂ diffraction peaks to a smaller angle. The substituted Ti⁴⁺ ions may have also diffused in crystal structure of BHFO.

Figure 3 exhibits the surface images of nanoparticles and films annealed at 500 °C. The observed nanoparticles are nearly spherical in shape with average diameter of 45 nm as depicted in Fig. 3(a). It can be clearly seen from Fig. 3(a) that nanoparticles are agglomerated and thus they were dispersed in an ultrasonic bath before composite thin film preparation. On the other side, the grains of pure TiO₂ film are polygonal and crack-free (see Fig. 3(b)) signifying the fact that annealing and densification has occurred to a satisfactory degree. The average grain size of this film is larger than that of nanoparticles and is found to be ~60 nm. The average diameter of nanoparticles and grain size of thin films were determined using Image J software.

Figure 3

FESEM images of (a) BHFO nanoparticles (b) TiO₂ film (c) T⁹⁵B⁵ film (d) T⁹⁰B¹⁰ film and (e) T⁸⁰B²⁰ film along with the films’ cross-section micrographs. The magnification for all the images is x50000.

The addition of only 5 mol.% BHFO in TiO₂ has introduced crack as well as diminished the uniformity and polygonality of the grains (see Fig. 3(c)). Further additions of 10 and 20 mol.% BHFO nanoparticles have totally changed the morphologies of thin films (see Fig. 3(d,e)). T⁹⁰B¹⁰ and T⁸⁰B²⁰ films have irregular polygonal grains and spherical nanoparticles which have made the surface rough. It has become difficult to identify TiO₂ and BHFO nanoparticles separately from Fig. 3(d,e). Indeed, T⁸⁰B²⁰ with 20 mol.% BHFO has larger number of porosities, cracks and rougher surface compared to T⁹⁰B¹⁰. The FESEM surface morphology of T⁸⁰B²⁰ film is consistent with the peak intensities (2θ = 22.54°, 32.14°, 39.48°, 46.185° and 57.34°) of XRD pattern (as depicted in Fig. 2) implying high amount of BHFO nanoparticles in the TiO₂ matrix. Moreover, the thickness of the films is increasing with increasing mol.% of BHFO nanoparticles. It can be seen from the insets of Fig. 3(b–e), the average thicknesses of TiO₂, T⁹⁵B⁵, T⁹⁰B¹⁰ and T⁸⁰B¹⁰ are 200, 244, 477, and 635 nm respectively. The cross sections of the films are also indicating the well-adherence of films with the glass substrate. Considering all these aforementioned observations it can be reasonably stated that there is no appreciable segregation of BHFO nanoparticles in the TiO₂ matrix.

Figure 4 depicts the absorbance spectra for thin films and nanoparticles. The absorbance of BHFO nanoparticles shown in the inset of Fig. 4 was obtained from diffused reflectance data using Kubelka-Munk conversion function³⁹. The diffused reflectance data was converted to Kubelka-Munk function given by the following equation:

$$F(R)=\frac{{(1-R)}^{2}}{2R}$$

(5)

where R is the diffused reflectance value. On the other hand the absorbance of the films was determined using Beer- Lambert law as follows^40,41:

$${\rm{A}}=log\frac{{(1-R)}^{2}}{T}$$

(6)

where A, T and R stand for the absorbance, transmittance and reflectance of the film.

Figure 4

UV-vis absorbance spectra of films. The inset shows the absorbance spectra of BHFO nanoparticles.

Nanoparticles exhibit extensive visible light absorbance and absorb larger wavelength than the synthesized films. The steep shape of the spectrum indicates that the visible light absorption has happened as a result of electronic transition from the valence band to conduction band (O²⁻ 2p→Fe³⁺ 3d) in BHFO lattice²³. This phenomenon will be discussed in detail in the following section. In contrast, the pure TiO₂ film does not start absorbing substantially until the incident wavelength falls below 370 nm (see Fig. 4). This is typical behavior of TiO₂ film¹⁹. It is noticeable from Fig. 4 that the absorption edge of composite thin films shows progressive shift between the pure TiO₂ thin film and BHFO nanoparticles. Even though the absorption edge of T⁹⁵B⁵ film does not shift much and still absorb shorter wavelength (∼380 nm) than two other composite films Fig. 4 clearly depicts that T⁹⁰B¹⁰ and T⁸⁰B²⁰ films show a redshift and absorb visible light. The absorption edges of T⁹⁰B¹⁰ and T⁸⁰B²⁰ films are 440 and 545 nm respectively (see Fig. 4). It has been established that the interfacial charge of TiO₂ gets affected by diffusion of foreign ions which in turn influences the optical properties^42,43. Indeed, the decrement of Ti³⁺ concentration in T⁸⁰B²⁰ film obtained from XPS analysis (see Table 1) suggests the possibility of A³⁺ ions diffusion in the crystal structure of TiO₂. Therefore, the bathochromic shift of T⁹⁰B¹⁰ and T⁸⁰B²⁰ films could be due to the interdiffusion of A³⁺/Ti⁴⁺ ions across the interfaces between BHFO and TiO₂. In fact, due to interdiffusion of ions the electrons of these two films could be excited from the partially filled d orbitals of Fe³⁺(rather than the lower-lying O p orbitals) to the Ti 3d orbitals with lower energy visible light¹.

In order to have better ideas on the optical properties of the films, their band gaps were determined as shown in Fig. 5. The band gap of BHFO nanoparticles obtained from [E *F(R)]² vs E plot depicted in Fig. 5(a)⁴⁴ is found to be 1.88 eV, which is notably smaller than the previously reported values for both undoped⁴⁴ and many doped BFO⁴⁵. The reduced band gap of BHFO could be attributed to the following reasons: (a) Ho³⁺ ions are likely to have minimal degree of hybridization for a stable electronic configuration (4f¹⁰ 5d⁰ 6s⁰) which in turn may lead to the formation of unique energy level in between Fe 3d and O 2p and thereby effective band gap of Ho doped BFO is diminished, (b) according to some previous investigations, changes in Fe- O bond length and Fe-O-Fe bond angle by cation doping play a pivotal role in modifying one-electron bandwidth (W) and thus band gap of BFO^23,46. This claim has been substantiated by the empirical formula relating W with bond length and angle:

$$W\approx \frac{cos\omega }{{d}_{Fe-O}^{3.5}}.$$

(7)

where ω stands for ½[π- (Fe-O-Fe] and ${d}_{Fe-O}$ is the Fe-O bond length⁴⁷. The relationship between band gap and W can be given as: ${E}_{g}={\rm{\Delta }}-W$ where Δ is the charge transfer energy⁴⁷. Generally, BFO crystallizes in a rhombohedral phase and bond length (Fe-O) of this phase is greater than that of orthorhombic phase²³. Since BHFO nanoparticles have orthorhombic phase and have smaller bond length, they have appreciably larger bandwidth (W) value than that of BFO which could reduce the effective band gap energy of BHFO.

Figure 5

(a) [E*F(R)]² vs energy, E plot to calculate band gap energy of nanoparticles (b) indirect and (c) direct Tauc’s plots demonstrating the band gap energies of pure and composite thin films.

The optical band gap of the films were calculated using the Tauc’s law⁴⁸ as given below:

$$(\alpha h\,v)=A^{\prime} \,{(E-{E}_{g})}^{n}$$

(8)

where A′ is a constant, E is the photon energy, E_g is the band gap energy, α is the absorption coefficient and n is the power factor of the optical transition mode in a semiconductor, i.e., direct transition (n = 1/2) or indirect transition (n = 2). The band gaps of the films were obtained by extrapolating the linear portion of of (αhν)² vs E plot for direct transition (see Fig. 5(c)) and (αhν)^1/2 vs E plot for indirect transition (see Fig. 5(b)). Since the indirect band gaps of the films are more related to their absorption edge values (see Fig. 4) they have been assigned to the respective composite thin films. Even both experimental results and theoretical modeling suggest that TiO₂ has a direct forbidden gap, which is almost degenerated with an indirect allowed transition. Therefore, the indirect allowed transition dominates in the optical absorption just above the absorption edge due to the weak strength of the direct forbidden transition⁴⁹.

The Tauc’s plots of composite thin films corresponding to indirect transition shown in Fig. 5 have only one linear region. However, the curve for pure TiO₂ has more than one linear regions which indicates that there is more than one optical transition occurring in this film. The first transition of pure film occurred at 3.4 eV is its effective band gap which agrees reasonably well with previous studies^35,50. The second indirect transition occurs at 2.1 eV and it is believed that this sort of transition is attributed to the presence of OVs in deposited TiO₂ film¹⁹.

The indirect band gaps of T⁹⁵B⁵, T⁹⁰B¹⁰ and T⁸⁰B²⁰ films are 3.13, 2.87 and 2.46 respectively (see Fig. 5). Thus, the optical transition energies of composites thin films monotonically decrease with increasing mol.% of BHFO nanoparticles. A perceptive explanation for the reduced band gap is related to conduction band of two different semiconductors. The hypothesis is that if the conduction band edge of the sensitizing material is higher than the conduction band edge of TiO₂, electrons can transfer from the smaller band gap material to the conduction band of TiO₂^18,19. According to some previous reports BFO has relatively high conduction band edge^1,51 which makes electrons in BFO nanoparticles to be transferred easily into the lower lying CB of TiO₂ through the interface. Thus it is required to determine the conduction band (CB) and valence band (VB) positions of both BHFO and TiO₂ to elucidate optical transition phenomenon. The CB potential of BHFO and TiO₂ at the point of zero charge can be calculated successfully by the following empirical equation^52,53.

$${E}_{CB}={\chi }_{S}-{E}^{e}-0.5{E}_{g}$$

(9)

where E_CB is the CB edge potential, E^e is the energy of free electrons on the hydrogen scale (∼4.5 eV), E_g is the band gap energy of the semiconductor, χ_s is the electronegativity of the semiconductor, which is the geometric mean of the electronegativity of the constituent atoms⁵³. The detail information on χ_s ihas been provided in the Supplementary Information. Plugging E^e, χ_s values of TiO₂ (5.81 eV) and BHFO (5.86 eV), and their corresponding E_g values into the above equations, E_CB of TiO₂ and BHFO stands as −0.39 and 0.42 eV respectively. The valence band edge potential E_VB can be obtained by the equation: E_VB = E_CB + E_g and the calculated E_VB of TiO₂ and BHFO are 3.01 and 2.3 eV respectively. Figure 6 depicts the calculated energy levels of TiO₂ and BHFO.

Figure 6

Schematic illustration for the calculated energy level diagram indicating the conduction and valence band potentials of TiO₂ film and BHFO nanoparticles.

It can be clearly seen from Fig. 6 that BHFO has relatively higher CB potential than that of TiO₂. As discussed earlier, the redshift shown by T⁹⁰B¹⁰ and T⁸⁰B²⁰ films is probably induced by Fe, Bi, Ho/Ti interdiffusion in the interfaces which could raise the CB edge potentials of BHFO and TiO₂ and hence lower the optical transition energy at the interface between BHFO and TiO₂. Such reduced band-gaps of T⁹⁰B¹⁰ and T⁸⁰B²⁰ could be beneficial for the efficient utilization of visible light for photocatalysis. Moreover from XPS analysis a higher amount of OH⁻ has been found in T⁸⁰B²⁰ which allows the generation of large amount of highly reactive hydroxyl radicals (${{\rm{OH}}}^{\bullet }$) through oxidation of OH⁻ by photo induced holes rendering the composite film highly reactive during photocatalysis.

From the Hall effect analysis, the Hall coefficient is found to be negative for all the thin films which indicates that they are n- type semiconductors. Generally, TiO₂ is an intrinsic n-type semiconductor⁸. The formation of the OVs and titanium interstitials are responsible for this sort of conductivity⁵⁴. TiO₂ is a sub-stoichiometric compound with excess titanium under standard conditions. This sub -stoichiometry is accommodated as OVs or titanium interstitials formation. The following reaction mechanisms can be considered for the formation of OVs in TiO₂ crystal⁵⁵:

$${O}_{O}^{x}\mathop{\to }\limits^{Ti{O}_{2}}\frac{1}{2}{O}_{2}(g)+{V}_{O}^{\bullet \bullet }+2{e}^{\text{'}}$$

(10)

$$2T{i}_{Ti}^{x}+2{e}^{\text{'}}=2T{i}_{Ti}^{\text{'}}$$

(11)

Combining Equations (4 and 5)

$$2T{i}_{Ti}^{x}+\,{O}_{O}^{x}=2T{i}_{Ti}^{\text{'}}+\frac{1}{2}{O}_{2}(g)+{V}_{O}^{\bullet \bullet }$$

(12)

As follows the n-type conductivity behavior found in TiO₂ implies Equation (6) to be the dominant step for providing two excess electrons. The carrier concentration (CC), resistivity and mobility of pure TiO₂ thin film deposited using non aqueous sol-gel method are found to be 2.0 × 10¹⁶ cm⁻³, 1550.8 Ω-cm and 208 cm²/V.s respectively. The CC of TiO₂ found in the literatures ranges from 1 × 10¹⁶ cm⁻³ to 1 × 10²⁰ cm^−3 56,57. Thus the measured value of CC is within accepted range. The measured mobility of the film is also consistent with the previous study which reported almost similar CC⁸. The resistivity of conductor and semiconductor normally varies from 10⁻³ to 10⁸ Ω-cm⁵⁸. Based on this range, pure TiO₂ film can be classified as semiconductor. However, the resistivity of this pure TiO₂ film is lower than prepared by many methods such as chemical bath deposition⁵⁸, spray pyrolysis⁵⁹ and DC magnetron sputtering technique²⁶.

Figure 7(a) depicts that 5 mol.% BHFO addition has significantly altered the CC, mobility and resistivity of the film. The resistivity of T⁹⁵B⁵ film is ∼2.33 orders of magnitude higher than pure film while the CC and mobility of this film are 62.5 and 2.6 times lower than those of pure one. Moreover, the charge carriers of T⁹⁵B⁵ film are dominated by free electrons at 5 mol.% BHFO addition, even though, BFO is reported to be a p-type semiconductor where holes are the majority carriers. Here, substitution of Bi³⁺ with Ho³⁺ should retain BHFO as p-type semiconductor^60,61. The determination of CC of BHFO was out of scope for the current study. Moreover, to the best of our knowledge there is no report on CC of BFO or on doped BFO^60,61. Since TiO₂ is the major phase its carrier will ultimately control the carrier concentration. The concentration of electrons may also be affected subtly by the diffused Fe ions. Since Fe ions have more than one valence states such as Fe²⁺ or Fe^3+ 62, some of diffused Fe³⁺ may have changed into Fe²⁺ by accepting free electrons of TiO₂ matrix and reduces total CC of T⁹⁵B⁵ slightly. Thus the total CC of the composite film decreases with the addition of BHFO.

Figure 7

(a) Resistivity, carrier concentration and mobility of films with different mol.% of BHFO (b) Conductivity (Ω-cm) vs. Electric Field, E (kV/cm) plot of BHFO nanoparticles.

The additions of 10 and 20 mol.% BHFO have not changed the electrical properties of T⁹⁰B¹⁰ and T⁸⁰B²⁰ films (see Fig. 7(a)) appreciably. The mobilities of T⁹⁰B¹⁰ and T⁸⁰B²⁰ films have decreased by a factor of 1.23 and 1.6 respectively and the CCs of these two films have been reduced by a factor of ∼4 .1 when compared to those of T⁹⁵B⁵ film. Similarly, the resistivities of T⁹⁰B¹⁰ and T⁸⁰B²⁰ films are 1.05 and 1.15 times higher than that of T⁹⁵B⁵ film. However, the charge carriers of T⁹⁰B¹⁰ and T⁸⁰B²⁰ films are still found to be dominated by free electrons. The aforestated variations in CC, mobility and resistivity point out that addition of higher mol.% of BHFO has little effect on electrical properties of composite films. The trivial reduction of CC in T⁹⁰B¹⁰ and T⁸⁰B²⁰ films primarily depends on diffused A³⁺ ions which substitute Ti⁴⁺ from the lattice sites of TiO₂ as discussed in XPS pattern analysis section. This substitution may lead to formation of holes compensating free electrons in TiO₂ and eventually reduces the films’ electron/Ti³⁺ concentration. However, further reduction in CC (electron hoping site concentration) is probably hindered by intergranular cracks and pore formations in the films (see Fig. 3(d,e)) as well as the direct contact interface areas between BHFO and TiO₂ are likely to become gradually saturated at higher BHFO nanoparticles content (10 and 20 mol.%) in the composite films. These phenomena could hamper further enhancement in migration of Ho³⁺, Fe³⁺ from BHFO to TiO₂ in T⁹⁰B¹⁰ and T⁸⁰B²⁰ restricting any marked alteration in overall carrier concentration of the system.

The increased resistivity of composite thin films is directly related to low conductivity of BHFO (see Fig. 7(b)). The conductivity of BHFO varies from 1 × 10⁻¹² to 5 × 10⁻¹² (Ω. cm)⁻¹ which is lower than that of previously reported doped BFO²³. The CC and mobility are also related to resistivity via the following equation⁶³:

$$\rho =\frac{1}{qn\mu }$$

(13)

where q is the electron charge, n is carrier concentration and μ is the mobility. The reduced CC and mobility of composite films could be ascribed to increased resistivity of films. Moreover, the poor morphology of the films may reduce mobility of films. Yasuno et al.⁶⁴ and Trinh et al.⁶⁵ observed that electrical properties of the film depend on surface roughness and porosity of film. These surface roughness and porosity increase the scattering of electrons and holes which in turn reduce their mean free path. From FESEM images it is clear that the pure film has a smoother surface than the composite films. With increasing mol.% of BHFO, the surface roughness and porosity of the films increase and charge carrier mean free path of the films decreases. Thus, the mobilities of the films decrease with reduced mean free path of electrons. The composite films with enhanced resistivity, and reduced CC and mobility will have minimum leakage current and reduced power losses in devices. Therefore, these films could be electrically reliable for resistors, sensors and memory devices.

Generally BFO nanoparticles exhibit G-type antiferromagnetism due to its cycloid spiral spin structure²³. However, an enhancement in magnetization by Ho-doping at Bi-site is evidenced in this study (see Fig. 8(a)). The saturation magnetization (Ms), coercivity (Hc) and magnetic susceptibility (χ_M) of BHFO nanoparticles are 24.33 emu/cm³, 260 Oe and 0.247 respectively. The density of BHFO is 8.40 g/cm³ which was calculated from unit cell parameters listed in Table S1. The magnetization unit of BHFO was converted from emu/gm to emu/cm³ by multiplying with 8.40 g/cm³. The bond lengths and bond angles of BHFO listed in Table S1 are quite different from the reported bond lengths and bond angles of pure BFO²³. The enhanced magnetization could be attributed to these changes in bond angle and bond length which modify the tilting angle of FeO₆ octahedron and thereby suppress spiral modulated spin stricture²³. Moreover, the substitution of non-magnetic Bi³⁺ ([Xe] 4f¹⁴ 5d¹⁰ 6 s² 6p⁰) with magnetic Ho³⁺ ([Xe] 4f¹⁰) having high magnetic moment (10.6 μ_B) could lead to improved magnetization.

Figure 8

Room temperature (a) M-H hysteresis loop for BHFO nanoparticles with inset showing its magnified M-H loop (b) M-H hysteresis loops for pure and composite films with inset showing magnified M-H loops of T⁹⁵B⁵, T⁹⁰B¹⁰ and T⁸⁰B²⁰ films.

Room temperature M-H curve of thin films were obtained after subtracting the contribution of glass substrate. TiO₂ film deposited on this substrate is diamagnetic (see Fig. 8(b)) and the diamagnetic susceptibility (χ_D′) is ∼−1.58 × 10⁻⁰³. Donor type defects (e.g. OVs) have been proposed to be a vital element for ferromagnetism in TiO₂^66,67 indicating that synthesized TiO₂ film may not have sufficient amount of OVs to induce magnetism. The other three composite thin films show no such diamagnetic behavior yet they display no complete ferromagnetic loop either. The composite thin films have ferromagnetic behavior accompanied with a diamagnetic component. The discernibly open hysteretic loop shown in the inset of Fig. 8(b) is signifying the presence of a ferromagnetic component. The actual saturation magnetization of all the composite thin films is likely to occur at low magnetic fields and all the films have diamagnetic components at higher magnetic fields. Figure 8(b) clearly depicts that saturation magnetization of the films increases with increasing mol.% of BHFO. T⁸⁰B²⁰ has higher magnetization and coercivity than two other composite films (see inset of Fig. 8(b)). Moreover, it has retained its ferromagnetism upto 11 kOe and has the lowest diamagnetic susceptibility (3.28 × 10⁻⁴). The diamagnetic susceptibilities of other two films are 1.01 × 10⁻⁰³ (T⁹⁵B⁵) and 6.045 × 10⁻⁴ (T⁹⁰B¹⁰). Previous investigations reveal that diffusion of magnetic ions into an oxide ceramic through interfaces aids in emergence of room temperature ferromagnetism^68,69. As conferred earlier, the interdiffusion of Fe³⁺, Ho ³⁺ and Ti⁴⁺ in the interfaces brings about a decrement of Ti³⁺ concentration in the crystal structure of TiO₂. The ferromagnetic exchange between the diffused magnetic ions (Fe³⁺ or Ho³⁺) and OVs could induce magnetism^68,70. The lacking of sufficient diffused magnetic ions to couple with OVs possibly leads to diamagnetism at higher magnetic field for the films. Since T⁹⁵B⁵ and T⁹⁰B¹⁰ have lower mol.% of BHFO compared to T⁸⁰B²⁰, they are expected to have smaller amount of diffused Fe³⁺ or Ho³⁺ ions in TiO₂ phase than that of T⁸⁰B²⁰. Therefore, the ferromagnetic exchange of T⁹⁵B⁵ and T⁹⁰B¹⁰ is weak due to small amount of diffused magnetic ions and thus diamagnetic susceptibility of these films is higher than that of T⁸⁰B²⁰. On the other hand, T⁸⁰B²⁰ film has higher amount of magnetic ions when compared with other two composite films. Hence, a stronger ferromagnetic exchange between magnetic ions and OVs is expected in T⁸⁰B²⁰ film leading to a lower diamagnetic susceptibility compared to the other two composite films. Indeed, composites thin film with further addition of BHFO nanoparticles (>20 mol.%) will have diamagnetic component. It can be seen from Fig. S5 that the reduction in diamagnetic susceptibility from TiO₂ film to T⁹⁵B⁵ film is prominent but further addition of BHFO nanoparticles does not bring such kind of change. In fact Fig. S5 depicts that diamagnetic susceptibility of T⁸⁰B²⁰ film is approaching saturation phenomenon conforming our explanation provided in electrical section. The presence of pores and cracks in films (see Fig. 3(d,e)) as well as possible saturation of total amount of direct contact interface areas between BHFO and TiO₂ not only affect the electrical properties but also the magnetic properties of T⁹⁰B¹⁰ and T⁸⁰B²⁰ films. Nevertheless, this finding suggests that incorporation of BHFO nanoparticles into TiO₂ matrix introduces magnetism in the final product which can be applied to thin films-based spintronics application.

Conclusion

In conclusion, novel BHFO/TiO₂ composite thin films were prepared for the first time by non-aqueous sol-gel method. The incorporation of BHFO nanoparticles improves magnetism and electrical properties of composite thin films, and also reduces their band gaps. The composite thin films exhibit a redshift in optical absorption edge and absorb visible light. Such behavior has been attributed to interdiffusion of A³⁺/Ti⁴⁺ ions through the interfaces between BHFO and TiO₂. Besides, interfacial microstructural defects and the direct contact areas between BHFO and TiO₂ in composite films control the migration of these ions through their interfaces which eventually determines the amount of CC in the films. Moreover, the ferromagnetic exchange between the diffused magnetic ions (Fe³⁺ or Ho³⁺) and OVs has been considered to influence the induction of magnetism in composite thin films.