Phase transition, interband electronic transitions and enhanced ferroelectric properties in Mn and Sm co-doped bismuth ferrite films

Abstract

Mn and Sm co-doped BiFeO₃ (BFO) polycrystalline films have been fabricated on F-doped SnO₂ (FTO) conductive glass substrate by a chemical solution deposition method. X-ray diffraction patterns and Raman spectra both imply a coexistence of rhombohedral and tetragonal phases in Mn–Sm co-doped BFO films. Scanning Electron Microscopy (SEM) images indicate more impact morphology with greater particle size in doped films. X-ray photoelectron spectroscopy (XPS) analysis suggests that high-valence Mn ions induced by Sm-doping lead to efficient decrease in oxygen vacancies, and thus improve leakage and polarization properties. Optical spectra analysis by using a Kubelka–Munk function and corresponding XPS valence band spectra analysis reveal that the optical band gap (E_g, varied from 2.38 eV to 2.26 eV) and Fermi level (E_F, varied from 1.18 eV to 1.01 eV) decrease continually with the introduction of Sm dopants, further confirming the decreased defects and vacancies in layer interfaces or grain boundaries. Consequently, larger saturation polarization and lower leakage current densities were obtained in Mn–Sm co-doped BFO films. Furthermore, possible energy band structure schematic proposed on the basis of above analysis shows that the shift in Urbach impurities energy level induced by ions dopants outsiders may be quite crucial in future photovoltaic application.

---

1. Introduction

BiFeO₃ (BFO) has been widely investigated for decades both in fundamental physics and promising application due to its intriguingly large spontaneous polarization, magnetic–electric coupling and anomalous photovoltaic effect, which also make it a potential candidate for future piezoelectric detectors or sensors, nonvolatile lead-free memory storage and novel photovoltaic devices.^1–3 For decades, great effort has been made such as structure engineering and ions doping to optimize both crystal structure and multiferroic properties of BFO.^4,5 Monocrystalline BFO films with admirable magnetoelectric properties have been deposited successfully by pulse laser deposition,^6,7 while such high costs in film preparation impede its industrialization. Therefore, much attention has been transferred to the polycrystalline films by chemical solution deposition, attempting to change its microstructure and improve its multiferroic or optical properties.^8–10

Furthermore, various research has been made to investigate both fundamental physics and practical applications of BFO polycrystalline thin films with site-engineering method because of the poor ferroelectric properties observed in pure BFO thin films due to the instability of phase structure and volatility of Bi element, while high performance in electrical properties including both leakage behaviour and ferroelectric properties have been confirmed in B-site Mn ions doped BFO thin films and A-site Sm ions doped BFO thin films respectively, probably resulting from phase structure transition, less defects and compress strains induced by their smaller ions radius than B-site Fe ions and A-site Bi ions respectively.^11–14 In our previous work, BFO films with 7.5% Mn ions replacement show the glaring ferroelectric remnant polarization in agreement with the reported 7% by S. K. Singh et al. in ref. 8. Moreover, Xu's research confirms the further non-centrosymmetry distortion by varying iron-oxygen octahedral and long-range ferroelectric order with Sm doping.¹⁵ Thus, we facilitate the Bi_1−xSm_xFe_0.925Mn_0.075O₃ (BSFM, x = 0, 0.025, 0.05, 0.075, and 0.10) films on F-doped SnO₂/glass (FTO) substrates by a solution deposition process and expect to obtain further improvement in its electric properties.

However, there still remains some controversies in the investigation of polycrystalline BFO films, including (I) ions doping leading to the microstructure transition: pure BFO crystal structure has been repeatedly considered as nearly-tetragonal phase structure or mixed tetragonal and rhombohedral phase due to its lattice mismatch with the substrate, the existence of morphotropic phase boundary and multilayer interface strain which is sensitive to film thickness.¹⁶ Few papers focused on the ions doping structure transition for it's a more puzzling and complicated process as it may introduce new unexpected impurities effect; (II) electronic energy band structure or inter-band electronic transitions: local electron orbits interactive hybridization and its coupling with certain electrons' high-spinning behavior are quite uncertain,¹⁷ while ions dopants outsiders probably result in the shifting of impurities energy lever or inter-band electronic transition energies and it's still unclear whether this behavior is favorable to practical applications or not.

All these controversies as the springboard of this paper inspire us to investigate the interplay between the crystalline structure, inter-band electronic transitions and physical properties in Mn–Sm co-doped BFO films.

2. Experimental

Pure BiFeO₃ and Bi_1−xSm_xFe_0.925Mn_0.075O₃ (BSFM, x = 0, 0.025, 0.05, 0.075, and 0.10) films were deposited on F-doped SnO₂/glass substrates by a solution deposition process.¹⁸ X-ray diffraction (XRD, X'pert PRO) and laser Raman spectrometer (HR800) were used to investigate the crystalline structure and phase transition in BSFM films. The morphology and thickness of these films were observed by scanning electron microscope (SEM, Quanta F250). The ferroelectric properties were measured by using an Aixacct TF Analyzer 2000, and the leakage current densities were characterized by Keithley 4200-SCS. The ionic valences were collected by X-ray photoelectron spectroscopy (XPS, AXIS ULtrabland). The room temperature Ultra-Violet-Visible (UV-Vis) reflectance spectra were measured by using a PerkinElmer spectrometer (Lambda 950).

3. Results and discussion

3.1 XRD patterns

Fig. 1 shows the XRD patterns for pure and Mn–Sm co-doped BFO films, revealing pure phases without any secondary phase in all the samples. All the diffraction peaks in the XRD patterns of pure BFO film can be easily indexed as a rhombohedral and distortion perovskite structure with a space group of R3c (PDF# 86-1518), evidenced by the noticeable doublet in around 32°, extra peak in 37° and splitting of the peak around 67°. However, none of these features maintains in the BSFM films thus suggesting an occurrence of incomplete structure transition possibly from rhombohedral phase (R-phase) to nearly-tetragonal phase (T-phase P4mm, referring to PDF#74-2459).¹⁹ Based on our previous work of Mn doping BFO films, a slight phase transition from rhombohedral phase to nearly-tetragonal phase has been confirmed and explained well due to the smaller radius replacement of B-site Fe ions. Furthermore, the smaller A-site Bi radius modification by Sm ions will possibly intensify the asymmetric distortion, so we mainly focused on the further modification of Sm ions to the microstructure of polycrystalline BFO films.

Fig. 1 XRD patterns of the pure BFO and BSFM_x (x = 0.025, 0.05, 0.075) films deposed on FTO/glass substrate by sol–gel process. Note that PDF#86-1518 card and PDF#74-2459 are used to clarify the phase transition from R3c space group to P4mm space group.

The elaborate fits of 32° peak for pure BFO and BSFM films are shown in Fig. 2. The hints in doublet fusion and broadening can be extracted easily from the peak-fitting–simulating parameters summarized in Table 1. These results further confirm the phase transition from R-phase to nearly T-phase, because T-phase (110) peak strengthens while R-phase (104) peak weakens shown by their increased peak area ratio from 0.67 to 1.53 and decreased interval from 0.274 to 0.248. Furthermore, these peaks broadening and fusion behaviour is probably derived from the introduction of Mn ions,²⁰ as well as the addition of Sm ions.

Fig. 2 XRD-peak-fitting–simulating analysis of 32° main peak including (104) and (110) sub-peaks of BFO and BSFM_x (x = 0.025, 0.05, 0.075) films. Note that the green line in the left represents for the (104) sub-peak from R-phase while blue line for the (110) sub-peak from T-phase.

Table 1 Unit cell parameters of BFO and BSFM_x (x = 0.025, 0.05, 0.075) films extracted from the refinements, peak-fitting–simulating analysis and calculation

Sample category	ac (Å)	cc (Å)	cc/ac	Local inhomogeneous distortion (cc−ac)/cc	(110) & (104) peak interval (θ)	(110) & (104) peak area ratio
Pure BFO	3.943(3)	3.947(4)	1.0010(1)	0.10(3)%	0.274	0.69
BSFM0.025	3.947(6)	3.955(6)	1.0020(2)	0.20(2)%	0.273	1.35
BSFM0.05	3.947(7)	3.955(8)	1.0020(5)	0.20(4)%	0.248	1.53
BSFM0.075	3.947(8)	3.957(9)	1.0025(6)	0.25(5)%	0.266	1.13

---

In order to establish the correlation between phase structure transition and polarization property, we carried out the Rietveld refinements for XRD data. The results and analysis reveal that the cubic-like lattice parameter a_c and c_c were affected by Sm doping concentration. For example, c_c/a_c ratio and local inhomogeneous distortion vary from 1.0010 to 1.0025 and from 0.10% to 0.25%, respectively. Therefore, the ferroelectric polarization that is directly proportional to the square-root of the distortion is believed to present significantly enhancement, which is confirmed in the later ferroelectric measurements.²¹ Above all, a complicated mixed phase structure including both R3c group and P4mm group was obtained in all Mn and Sm co-doped films, which is consequently confirmed by the laser Raman spectra in the next section. Simultaneously, this mixed phase structures will probably induce a more complicated electronic band structure and tempting polarization or other physical performance.²²

3.2 Raman spectra

Further analysis of micro-structure transition was carried out by Raman spectra with 631 nm laser excitation. Standard rhombohedral R3c structure of pure BFO film can be certificated from Fig. 3 due to the appearance of 13 typical Raman active modes (4A₁ + 9E) corresponding to R3c. However, only 8 Raman models can be fitted in BSFM films, which looks like much closer to P4mm structure (3A₁ + B + 4E).²³ In order to obtain the exact peak position, each measured spectrum for BSFM films was properly fitted by decomposing the fitted curves into individual Lorentzian components (Raman spectra of pure BFO film and BSFM (x = 0.05) film are selected to clearly exhibit the fitting procedure, as shown in Fig. 3). Corresponding Raman shift values were summarized in the Table 2. Obviously, the quantitative values of pure BFO film significantly turn in a consistent performance with the R3c group in ref. 24, while BSFM films show dramatic changes in amount, position, and intensity of the corresponding peaks.^24,25 It should be mentioned here that the accuracy of the equipment that limits the veracity of the ultra-low wavenumber measurement below 100 cm⁻¹ and this part of fuzziness should not affect the whole variation tendency between 100 cm⁻¹ to 800 cm⁻¹.

Fig. 3 Raman spectra and Raman-peak-fitting–simulating analysis of the BFO and BSFM_x (x = 0.05) films under excitation 631 nm at room temperature. The red dashed lines show the positions of Raman-active photon modes of BSFM_x (x = 0.05). Note that blank lines are the raw intensity and colored lines are the fitting components.

Table 2 Raman active modes of R3c and P4mm structure from references and of as-prepared BFO and BSFM_x (x = 0.025, 0.05, 0.075) films

R3c structure	Raman active models
	E-1	A1-1	A1-2	A1-3	E-2	E-3	E-5	E-6	E-7	E-8	E-9	B (P4mm structure)
R3c : H (ref. 18)	76.0	140	172	214	262	275	345	369	470	521	613
Pure BFO	75.9	141	172.5	219.5	262.2	273.7	347.9	370.8	470.6	521.0	611.8
BSFM0.025	73.3	135.1	169.2		247.8			379.6	479.7	543.0		628.9
BSFM0.050	73.5	134.6	169.6		247.5			382.2	479.6	542.3		629.0
BSFM0.075	73.8	134.3	169.4		249.0			385.7	481.5	542.9		628.5

---

P4mm structure	Raman active models
	A1-1	A1-3	E-2	E-3	E-4	B
P4mm (ref. 19)	147	227	266	273	369	691

---

The first-principles calculation²⁶ suggests that Bi atoms only participated in low-frequency modes up to 167 cm⁻¹ and the motion of oxygen atoms dominated the modes above 262 cm⁻¹, whereas Fe atoms were mainly involved in modes between 152 and 262 cm⁻¹ with possible contribution to higher frequency modes. Correspondingly, the two characteristic modes of A₁-1 mode at ∼140 cm⁻¹ and E mode at ∼75 cm⁻¹ should be governed by Bi–O covalent bonds, which control the dielectric constant and the ferroelectric phase of BiFeO₃.^27,28 As seen in Fig. 3, E mode at ∼75 cm⁻¹ presents dramatic increase in relative intensity for all BSFM films compared to that of pure BFO, suggesting a significant change in ferroelectric properties. Especially, a B-mode-like peak appears at the position of about 628 cm⁻¹, as well as some E modes of high frequency above 350 cm⁻¹ (asterisked E peak in Raman spectra, while the asterisked A₁-3 mode is probably merged with E models near 250 cm⁻¹) disappeared and the rest Raman modes undergo high frequency shifting, strongly implying the coexistence of R3c and P4mm phases as analysed above in XRD patterns. More details about transformation in electronic band structures will be discussed deeply in subsequent XPS analysis and reflectance spectra sections.

3.3 Morphology

In Fig. 4, BSFM films with an approximately thickness of ∼300 nm were well deposited on FTO glass substrate, having distinct interfaces between different layers as shown in SEM cross-section image. The thickness of BFO ultra-thin BFO films below 100 nm is directly proportional to the ferroelectric properties which results from the interface strain due to the crystal mismatch while the enhanced polarization properties of our polycrystalline specimen are closely related with the grain size, grain distribution and porosity due to their similar thickness around 300 nm with the error of 3%. It's easy to control the thickness of the BFO film above 300 nm by sol–gel coating method and this factor may not be the dominant one in our so-called “thick” films due to their similar section morphology around 300 nm.²⁹

Fig. 4 SEM images of BFO film and BSFM_x (x = 0.025, 0.05, 0.075) films and cross-section structure of BSFM x = 0.05 film. Note that we obtain a same thickness of ∼300 nm in all films prepared at the same condition.

Since grain boundary and size are of a great significance in ferroelectric properties and leakage conduction, pure BFO film shows smaller grain size ranging from 80–180 nm with numbers cracks or pores which may easily form the traps of charge at crystal interface and boundaries, thus causing the undesirable ferroelectric properties in practical application.³⁰ A small quantity of Mn dopants may expand the particles and improve the leakage property as we reported before in ref. 18. Actually, average grain size doubled to be over 400 nm with 2.5% Sm-doping concentration and extensive small grain appeared to merge with its surroundings, then this grain merging phenomenon persistently happened until almost all grains came to be contiguous over several micrometres like a plane as Sm doping concentration came to be over 7.5% doping concentration. Although BSFM (x = 0.025, 0.05) film shows some pinholes and cracks but it has a better ferroelectric and leakage property mainly for the grain merging resulted in the bigger but less cracks and pinholes than pure BFO film which means the less defects at the boundary. On the other hand, the less but lager grains signify the better uniformity and orientation on account of the less grain boundary in polycrystalline BFO films based on the less oxygen defects by Sm-doping. However, a large number of new small grains with the grain size below 100 nm gradually grew and wrapped at the large particles boundaries in the surface of BSFM x = 0.075 film and somehow lower its polarization property. It should be mentioned that grain size and surface defects probably play an important role in leakage and optical behavior as discussed later.

3.4 Leakage properties

Fig. 5 show the leakage current densities at the applied electric field from −333 kV cm⁻¹ to 333 kV cm⁻¹, of which BSFM films lower the leakage current density about one order of magnitude than that of pure BFO. It should be attributed to the limitation to oxygen defects by non-volatile Sm ions and high-valence Mn⁴⁺ ions on basis of the possible chemical equation as follow,³¹

Fig. 5 Leakage current density J vs. electric field E of pure BFO and BSFM_x (x = 0.025, 0.05, 0.075) films measured at room temperature.

As seen in the above equation, oxygen defects were easily generated due to the Bi volatilization during the preparation of pure BFO. The introduction of non-volatile Sm will combine oxygen ions tightly to form Sm–O bond, and thus restrain the production of oxygen defects. Furthermore, the introduction of Mn²⁺ ions may probably react with oxidizing oxygen vacancy and consequently lower the density of oxygen defects by generating Mn⁴⁺ ions. In other words, more O-bonds and high valence Mn ions (Mn⁴⁺) will decrease the leakage current density of BSFM films compared to pure BFO. What's more, all these process can be demonstrated in the XPS analysis of O ions and Mn ions section later.

3.5 Ferroelectric properties

In Fig. 6, ferroelectric measurements were carried out at the frequency of 1 kHz at room temperature to obtain hysteresis loops and polarization switching current of BFO, BFM_0.075 (BiFe_0.925Mn_0.075O₃) film and BSFM (x = 0.025, 0.05, 0.075) films. During the ferroelectric measurement, we gradually increased the applied electric field to the critical breakdown value to obtain the maximum withstand capability. So the ferroelectric hysteresis loops were displayed at their respective maximum field corresponding to the varied resistance character. Saturation polarization values of ∼39, 62, 73, 52 μC cm⁻² were observed (see Table 3) with increasing Sm dopants at the applied field within 333 kV cm⁻¹, while pure BFO performed quite poor ferroelectric behavior due to its severe leakage behavior³² induced by oxygen defects and abominable morphology. It is clear that the best ferroelectric performance was obtained in the 5% Sm doped BFO film. Moreover, obvious polarization switching behavior shown in Fig. 6(b) was observed in all the BSFM (x = 0.025, 0.05, 0.075) films with the slight increase of the peak current densities. From the leakage analysis in section 3.4, BSFM x = 0.05 film shows the similar leakage properties with the other BSFM films, and it is also demonstrated by the similar diagonal slope of polarization switching behavior in Fig. 6(b), revealing the lager P_r value of BSFM x = 0.05 film mainly results from the different effects of the varied Sm doping concentration, including the change of grain sizes, oxygen defects densities and crystal distortion rather than leakage current compensation. Otherwise, the good performance in ferroelectric properties should be benefited from the lager grain size and the incomplete phase structure transition from R-phase to T-phase by introducing a small quantity of Sm dopants. We can reasonably assume from the analysis above that the crystal distortion with lager c_c/a_c ratio will probably cause large polarization due to enhanced polarization in our near-tetragonal phase specimen the reported lager c_c/a_c ratio of ∼1.25 in tetragonal-phase BFO structure.¹⁶ Nevertheless, excess Sm doping concentration (over 7.5% in this case) will probably suppress polarization switching process to some content because its particular pinning effect becomes dominant.³³ Therefore, enhanced polarization properties have been accessed by Mn and Sm co-doping in this case, forecasting a good future application in low-cost solution-deposited BFO films.

Fig. 6 Ferroelectric hysteresis loops (a) and current density hysteresis loops (b) of BFO, BFM_0.075 (BiFe_0.075Mn_0.925O₃) and BSFM_x (x = 0.025, 0.05, 0.075) films measured at the applied alternating electric field of 1 kHz at room temperature.

Table 3 Key parameters of ferroelectric polarizations for pure BFO, BFM_0.075 (BiFe_0.075Mn_0.925O₃) and BSFM_x (x = 0.025, 0.05, 0.075) films, their polarization reverse electric fields and corresponding c_c/a_c value

Sample category	PS (μC cm^−2)	Pr (μC cm^−2)	EC (kV cm^−1)	EJ max (kV cm^−1)	cc/ac (arb. units)
BFO	—	∼0	∼20	—	1.0010(1)
BFM	39	35	106	74	—
BSFM0.025	62	53	130	83	1.0020(2)
BSFM0.05	73	64	187	105	1.0020(5)
BSFM0.075	52	40	115	67	1.0025(6)

---

3.6 XPS analysis

In order to explore electronic structure evolution and co-doping investigated by XPS spectroscopy. Fig. 7 and 8 show further analysis of O 1s and Mn 2p_3/2 orbits of BFO film and BSFM (x = 0.025, 0.05, 0.075) films and the corresponding peak-fitting parameters have been concluded in Tables 4 and 5, respectively. Firstly, the fitting result of Fe 2p_3/2 orbits display singlet look (not shown here), implying only Fe³⁺ ions exist in our BFO film and BSFM films, and hence some other factors must be dominant in the leakage mechanism like Bi³⁺ ions volatilization or oxygen vacancy generation in film preparing process.³⁴ As shown in Fig. 7, the O 1s main peak at ∼529.5 eV at lower binding energy are associated with the O–X bonds (X represents for Bi, Sm, Mn, and Fe), while another one at higher binding energy of ∼531.5 eV are associated with the O defects or vacancies. As seen in Table 4, it is easy to clarify by the increased proportions of O–X bonds from 2.32 to 6.60 in the whole O 1s peaks, implying the formation of more non-volatilize stable O–Sm bonds suppress the generation of the O vacancies serving as the channel for electrons migration, simultaneously leading to the band energy and electronic structure transition.³⁵

Fig. 7 O 1s orbits X-ray photoelectron spectra of BFO and BSFM_x (x = 0.025, 0.05, 0.075) films and XPS-peak-fitting–simulating analysis of O defect peak (green line) at ∼529.2 eV and O–X bond peak (blue line) at ∼531.4 eV.

Fig. 8 Mn 2p_3/2 orbit X-ray photoelectron spectra of BSFM_x (x = 0.025, 0.05, 0.075) films and XPS-peak-fitting–simulating analysis of Mn²⁺ peak (green line) at ∼641.2 eV and Mn⁴⁺ peak (blue line) at ∼531.4 eV.

Table 4 Primary parameters of XPS-peak-fitting–simulating analysis for O 1s orbits of BFO and BSFM_x (x = 0.025, 0.05, 0.075) films

O 1s orbit	Chemical bonding	Binding energy (eV)	Peak area (arb. units)	Peak area ratio
Pure BFO	O–X bond	529.3	60 833	2.32
	Defect O	531.5	26 259
BSFM0.025	O–X bond	529.2	127 045	3.76
	Defect O	531.3	33 809
BSFM0.05	O–X bond	529.2	105 578	4.58
	Defect O	531.4	23 061
BSFM0.075	O–X bond	529.3	96 380	6.60
	Defect O	531.4	14 616

---

Table 5 Main parameters of XPS-peak-fitting–simulating analysis for Mn 2p_3/2 orbits of BSFM_x (x = 0.025, 0.05, 0.075) films

Mn 2p3/2 orbit	Chemical valence	Binding energy (eV)	Peak area (arb. units)	Peak area ratio
BSFM0.025	+4	642.7	4409	1.23
	+2	641.1	3575
BSFM0.05	+4	642.7	4460	1.26
	+2	641.1	3529
BSFM0.075	+4	642.3	3202	1.57
	+2	641.3	2029

---

However, the dividing peaks in Mn 2p_3/2 XPS spectra corresponding to ∼642.7 eV and ∼641.3 eV peaks display the variable valences of Mn ions considered as Mn⁴⁺ and Mn²⁺ ions respectively. As seen in Table 5, their area ratio varied from 0.24 to 1.57 shows that more Mn⁴⁺ ions appeared and less oxygen vacancies induced with the introduction of Sm dopants, confirming the hypothesis of chemical equations above. Therefore, improved ferroelectric polarization properties may be originated from synergistic effects including the following facts, such as greater lattice distortion, phase transition from rhombohedral to tetragonal phase, less crystal interfaces and smooth morphology, greater grain size, less oxygen vacancies from the limitation of Sm ions, high-valence Mn ions and lower leakage behavior.

3.7 Optical properties

Optical spectroscopy is a widely used to probe electronic band structure in solids, especially it could reveal rich physics in complex oxides. More recently, it has attracted much attention to study the electronic band structure and in-gap states in BiFeO₃ compound, since it will reveal the interplay between charge, structure and physical properties in this prototype multiferroic system.

The reflectance spectra collected from UV-Vis spectroscopy was recorded over the range from 200–1000 nm in comparison with the Kubelka–Munk (K–M) models below,^35,36

where R represents for the reflectivity. Reflectance spectra other than transmittance spectra was used here to derive the band gap due to the some pollutant attached to the back of the glass substrate during the coating process, which may affect the veracity of transmittance data. The direct band gap approach of BFO and BSFM (x = 0.025, 0.05, 0.075) films could be estimated to be 2.38 eV, 2.29 eV, 2.28 eV and 2.26 eV respectively, by extrapolating (F(R) × E)² to zero of photon energy (in Fig. 9 inset (b)) and the highly noticeable variation of an absorption edge at around 428 nm for all specimens is corresponding to the charge-transition band at ∼2.9 eV (CT band shown in Fig. 9), indicating light absorption within visible range.

Fig. 9 Kubelka–Munk functions, F(R) vs. photon energy hν of BFO and BSFM_x (x = 0.025, 0.05, 0.075) films including p–d charge-transfer at ∼2.9 eV and d–d band absorption at ∼2.45 eV. Inset (a) shows two other absorption onsets in low energy band attributed to ⁶A_1g–⁴T_1g and ⁶A_1g–⁴T_2g crystal-field transitions, respectively. Inset (b) is the energy band gap linearly extrapolated from the low-energy part of (F(R) × hν)² to energy axis. Inset (c) displays Urbach energy of BFO and BSFM_x (x = 0.025, 0.05, 0.075) by calculating the slope of reciprocal extrapolated from liner part of ln(F(R)) vs. photon energy hν in low energy side. Note that reflectance (R) is derived from room-temperature UV-Vis reflectance spectra.

In fact, the interatomic transition and coupling effect³⁷ between electron charge (O 2p, Fe 3d, Bi 6s, Bi 5p levels) and absorption bands (d–d bands transition) due to the electron spin of Fe³⁺ (3d⁵ high-spin behavior) local lattice field in BiFeO₃ system make its local electronic band structure more complicated like the schematic shown in Fig. 10. By considering the C_3v local symmetry of Fe³⁺ ions in BiFeO₃ and using the correlation between group and subgroup analysis for the symmetry breaking from O_h to C_3v, the typical t³_2ge²_g high-spin configuration of Fe³⁺ ions in the cubic octahedral environment transforms into an a¹₁e²e² electronic configuration in the rhombohedral environment.^21,38

Fig. 10 Schematic energy splitting diagram of Fe³⁺-d⁵ orbits and further t²g³ orbits indicating the crystal-field effect with lower symmetry from cubic octahedral O_h to rhombohedral C_3V environment.

Despite the intense peak for the charge-transition band, two extra absorption onsets at low energy below 2.0 eV corresponding to ⁶A_1g–⁴T_1g and ⁶A_1g–⁴T_2g crystal-field d–d transitions have attract much attention because they are the corresponding splitting level. It is believed that these two crystal-field transitions of Fe³⁺ (3d⁵) are highly sensitive to slight distortions of octahedron FeO₆. Both the transition energy and oscillator strength strongly depend on the Fe–O bond length, site symmetry and Fe–O–Fe exchange interaction.³⁹ The foregoing XRD and Raman results reveal an obvious occurrence of phase transition and distortions in FeO₆ octahedron induced by Sm–Mn co-doping. Apparently, these changes should be detected by optical spectroscopy.

As shown in inset (a) of Fig. 9, it's easy to observe two extra absorption onsets at 1.47, 1.39, 1.43, 1.39 eV and 1.77, 1.64, 1.70, 1.65 eV for BFO and BSFM films (x = 0.025, 0.05, 0.075) respectively, corresponding to ⁶A_1g–⁴T_1g and ⁶A_1g–⁴T_2g crystal-field transitions. More interestingly, both two transitions energies exhibit apparent red-shift behavior in all doped samples. Based on the modified Tanabe–Sugano diagrams,⁴⁰ this red-shift in the two crystal-field transition energies should be attributed to increase in the e–t₂ crystal-field splitting Δ, induced by the smaller size Sm ions substitution of Bi-site in BiFeO₃. Furthermore, such a red-shift behavior reflect significant change in the local structure of Fe³⁺ in BiFeO₃ system, which indeed affect its physical properties.

Fermi level (E_F) of BFO and BSFM (x = 0.05) films can be decided by XPS valence spectra from Fig. 11(a) and (b). It could be observed that E_F is lower than the middle level of the band gap in BSFM (x = 0.05) film, because the dominating Mn⁴⁺ ions are kind of accepter dopants. Urbach energy E_U was determined through ln(F(R)) = hν/E_U relationship, varying from 0.1397, 0.1192, 0.1190, 0.2060 eV (from Fig. 9 inset (c)) below the conduction band (E_C) bottom of BFO and BSFM films (x = 0.025, 0.05, 0.075) respectively. It is known that Urbach energy E_U is closely related with the defects and strains.⁴¹ The lower Urbach energy value means less oxygen defects and impurities as the unexpected absorption centres existed in BSFM films (except BSFM x = 0.075), which is in good agreement with the foregoing XRD, Raman and SEM analysis. Based on above results, the electronic band structures for BFO and BSFM (x = 0.05) films were schematically proposed in Fig. 11(c) and (d) and their corresponding parameters have been listed in Table 6.

Fig. 11 X-ray photoelectron spectra of valence band of BFO (a) and BSFM_x (x = 0.05) film (b). The intersection plot of two linear fitting parts in their low-energy region are the Fermi-level energy of ∼1.18 eV and ∼1.01 eV respectively. Based on the reflectance spectra and XPS of valence band analysis, splitting energy levels of ⁶A_1g–⁴T_1g and ⁶A_1g–⁴T_2g transitions and impurities energy level E_U are schematically proposed as electronic energy band structure diagram (c) and (d).

Table 6 Corresponding parameters of local absorption level ⁴T_2g, splitting energy level ⁶A_1g–⁴T_1g and ⁶A_1g–⁴T_2g, absorption onset energy E_g and Urbach energy E_U for BFO and BSFM_x (x = 0.05) films respectively

Sample	^4T2g (eV)	^6A1g–^4T1g (eV)	^6A1g–^4T2g (eV)	Eg (eV)	EF (eV)	EU (eV)
Pure BFO	2.88	1.47	1.77	2.38	1.18	0.139(7)
BSFM0.05	2.88	1.43	1.70	2.29	1.01	0.119(0)

---

Briefly, Sm and Mn co-doped in BiFeO₃ induced remarkable distortions in FeO₆ and an incomplete phase transition evidenced by XRD and Raman spectra analysis, and thus led to significant changes in the local structure of Fe³⁺, which could be clearly reflected by a red-shift behavior in the crystal-field transition energies. Moreover, the lower Urbach energy value means less oxygen defects and impurities as the unexpected absorption centres existed in BSFM films. From a macroscopic point of view, these microstructures changes ultimately resulted in the improved leakage properties and the enhanced ferroelectric properties.

4. Conclusions

Pure BFO and BSFM films have been well fabricated on FTO conductive glass substrate by a solution deposition method. Microstructure analysis in XRD patterns and Raman spectra imply a coexistence of rhombohedral and tetragonal phases. It's clear to observe in XRD patterns that all multimodal peaks show a merging behaviour including R-mode (R, rhombohedral) peaks weakens and T-mode (T, tetragonal) peaks strengthen process. Moreover, amount of Raman active modes reducing from 13 (4A₁ + 9E) to 8 modes (3A₁ + B + 4E) confirms the incomplete structure phase transition. Scanning electron microscope images indicate more impact morphology by Sm-doping with greater particle size. Higher saturation polarization and lower leakage current density were obtained through the introduction of Sm and Mn. XPS analysis suggests that Sm-doping and high-valence Mn ions transition are beneficial to decrease oxygen vacancy density and improve polarization or leakage properties. Note that the direct optical band gap (E_g, from 2.38 eV to 2.26 eV) and Fermi level (E_F, from 1.18 eV to 1.01 eV) decrease with increasing Sm dopants from UV-Vis reflectance spectra by using Kubelka–Munk functions and corresponding XPS valence band spectra, confirming the decreased defects or vacancies in layer interfaces or grain boundaries by Sm modification. Subsequent energy band structure schematic was proposed with Urbach impurities energy level shifting from 0.14 eV to 0.12 eV below the conduction band bottom, indicating less absorption and recombination centres in doped BiFeO₃ film, which may be quite crucial in future photovoltaic research.

Acknowledgements

Financial support by the National Natural Science Foundation of China (Grant No. 51272204) is gratefully acknowledged. J. Wei wants to thank the China Scholarship Council (CSC) for funding his stay in France. The authors also thank Ms Dai and Mr Ma for their help in SEM and EDS analysis at International Centre for Dielectric Research (ICDR), Xi'an Jiaotong University, Xi'an, China.

Notes and references

1. R. Ramesh and E. Al, et al., Science, 2003, 299, 1719–1722.
2. T. J. Park, S. Sambasivan, D. A. Fischer, W. S. Yoon, J. A. Misewich and S. S. Wong, J. Phys. Chem. C, 2008, 112, 10359–10369.
3. J. Wu, J. Wang, D. Xiao and J. Zhu, ACS Appl. Mater. Interfaces, 2011, 3, 3261–3263.
4. P. Suresh and S. Srinath, J. Appl. Phys., 2013, 113, 17D920.
5. T. Kawae, Y. Terauchi, H. Tsuda, M. Kumeda and A. Morimoto, Appl. Phys. Lett., 2009, 94, 112904.
6. K. Y. Yun, D. Ricinschi, T. Kanashima, M. Noda and M. Okuyama, Jpn. J. Appl. Phys., 2004, 43, 647.
7. S. Y. Yang, F. Zavaliche, L. Mohaddes-Ardabili and V. Vaithyanathan, Appl. Phys. Lett., 2005, 87, 102903.
8. S. K. Singh, H. Ishiwara, K. Sato and K. Maruyama, J. Appl. Phys., 2007, 102, 094109.
9. S. K. Singh, H. Ishiwara and K. Maruyama, Appl. Phys. Lett., 2006, 88, 262908.
10. Y. Guo, B. Guo, W. Dong, H. Li and H. Liu, Nanotechnology, 2013, 24, 86–94.
11. S. Singh, C. Tomy, T. Era, M. Itoh and H. Ishiwara, J. Appl. Phys., 2012, 111, 102801.
12. J. Luo, S. Lin, Y. Zheng and B. Wang, Appl. Phys. Lett., 2012, 101, 062902.
13. M. Kumar and K. Yadav, Appl. Phys. Lett., 2007, 91, 242901.
14. C.-J. Cheng, D. Kan, S.-H. Lim, W. McKenzie, P. Munroe, L. Salamanca-Riba, R. Withers, I. Takeuchi and V. Nagarajan, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 80, 014109.
15. X. Xu, T. Guoqiang, R. Huijun and X. Ao, Ceram. Int., 2013, 39, 6223–6228.
16. D. Mazumdar, V. Shelke, M. Iliev, S. Jesse, A. Kumar, S. V. Kalinin, A. P. Baddorf and A. Gupta, Nano Lett., 2010, 10, 2555–2561.
17. P. S. V. Mocherla, C. Karthik, R. Ubic, M. S. Ramachandra Rao and C. Sudakar, Appl. Phys. Lett., 2013, 103, 022910–022915.
18. Y. Liu, J. Wei, Y. Liu, X. Bai, P. Shi, S. Mao, X. Zhang, C. Li and B. Dkhil, J. Mater. Sci.: Mater. Electron., 2016, 27, 3095–3102.
19. Z. Wen, X. Shen, D. Wu, Q. Xu, J. Wang and A. Li, Solid State Commun., 2010, 150, 2081–2084.
20. J. Z. Huang, Y. Shen, M. Li and C. W. Nan, J. Appl. Phys., 2011, 110, 094106.
21. X. Bai, J. Wei, B. Tian, Y. Liu, T. Reiss, N. Guiblin, P. Gemeiner, B. Dkhil and I. C. Infante, J. Phys. Chem. C, 2016, 120, 3595–3601.
22. M. K. Singh, S. Ryu and H. M. Jang, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 72, 2101.
23. H. M. Tütüncü and G. P. Srivastava, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 78, 1879–1882.
24. Y. Yang, J. Y. Sun, K. Zhu, Y. L. Liu and L. Wan, J. Appl. Phys., 2008, 103, 093532–093535.
25. M. N. Iliev, M. V. Abrashev, D. Mazumdar, V. Shelke and A. Gupta, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 175–227.
26. P. Hermet, M. Goffinet, J. Kreisel and P. Ghosez, Phys. Rev. B: Condens. Matter Mater. Phys., 2007, 75, 220102.
27. M. K. Singh and R. S. Katiyar, J. Appl. Phys., 2011, 109, 07D916.
28. R. Haumont, J. Kreisel, P. Bouvier and F. Hippert, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 73, 2101.
29. Y. Chu, T. Zhao, M. Cruz, Q. Zhan, P. Yang, L. Martin, M. Huijben, C. Yang, F. Zavaliche and H. Zheng, Appl. Phys. Lett., 2007, 90, 252906.
30. K. Yin, M. Li, Y. Liu and C. He, Appl. Phys. Lett., 2010, 97, 042101–042103.
31. G. Dong, G. Tan, Y. Luo, W. Liu, A. Xia and H. Ren, Appl. Surf. Sci., 2014, 305, 55–61.
32. H. Yang, M. Jain, N. A. Suvorova, H. Zhou, H. M. Luo, D. M. Feldmann, P. C. Dowden, R. F. Depaula, S. R. Foltyn and Q. X. Jia, Appl. Phys. Lett., 2007, 91, 072911–072913.
33. S. M. Wu, S. A. Cybart, D. Yi, J. M. Parker, R. Ramesh and R. C. Dynes, Phys. Rev. Lett., 2013, 110, 640–647.
34. G. Catalan and J. F. Scott, Adv. Mater., 2009, 21, 2463–2485.
35. Q. Ke, X. Lou, Y. Wang and J. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 175–227.
36. B. Ramachandran, A. Dixit, R. Naik, G. Lawes and M. S. R. Rao, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 1558–1564.
37. F. Gao, X. Y. Chen, K. B. Yin, S. Dong, Z. F. Ren, F. Yuan, T. Yu, Z. G. Zou and J. M. Liu, Adv. Mater., 2007, 19, 2889–2892.
38. J. B. Neaton, C. Ederer, U. V. Waghmare, N. A. Spaldin and K. M. Rabe, Phys. Rev. B: Condens. Matter Mater. Phys., 2004, 71, 4113.
39. R. G. Burns, Mineralogical applications of crystal field theory, Cambridge University Press, 1993.
40. S. Gómezsalces, F. Aguado, F. Rodríguez, R. Valiente, J. González, R. Haumont and J. Kreisel, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 85, 98–102.
41. K. Jiang, J. J. Zhu, J. D. Wu, J. Sun, Z. G. Hu and J. H. Chu, ACS Appl. Mater. Interfaces, 2011, 3, 4844–4852.

---