Enhanced optical and piezoelectric characteristics of transparent Ni-doped BiFeO₃ thin films on a glass substrate

Abstract

Very thin and highly transparent BiFeO₃ films are attractive owing to their potential application as ferroelectric photovoltaic devices. Here, we demonstrate enhancements in optical and piezoelectric properties that occur upon very low levels of Ni doping in solution-processed BiFeO₃ thin films (thickness less than 200 nm). Doping only with 0.5 mol% Ni reduces the optical band gap from 2.83 to 2.78 eV and increases the piezoelectric coefficient from 15.4 to 28.0 pm V⁻¹. These improvements are attributed to changes in the morphotropic phase boundary, oxygen-related defects, and crystallinity, which are driven by the low levels of Ni doping. For example, the increased piezoelectric coefficient upon Ni doping is attributed to a movement toward the morphotropic phase boundary and the enhanced crystallinity of the perovskite phase.

---

1. Introduction

BiFeO₃ (BFO) is a stable ferroelectric material that simultaneously possesses room-temperature ferroelectric and magnetic order.^1–3 Photovoltaic and resistive-switching behavior has also been reported.⁴ BFO films have a higher remnant polarization and a lower energy band gap compared with conventional ferroelectric materials such as Pb(Ti,Zr)O₃.^4,5 The depolarization field in the film provides a driving force to separate excited electrons and holes. Therefore, an ultrathin BFO film having strong self-polarization and high carrier density is desirable for better charge transport and polarity for photovoltaic applications. Most BFO thin films studied heretofore have thickness greater than 500 nm.^6–8 Recently, studies on optical and photovoltaic properties in BFO films on fluorine-doped tin oxide (FTO)–glass substrates have been reported.^9–11 However, studies that deal with both optical and ferroelectric or piezoelectric properties of thin BFO films optimized on FTO–glass substrates are rare.

In this study, we investigate the effect of Ni doping on the structural, optical, and piezoelectric properties of BFO thin films prepared by solution deposition. To keep the transmittance level above 80% in the visible wavelength region, these films needs to be relatively thin (∼200 nm). A few reports exist that discuss the effects of Ni substitution in BiFe_1−xNi_xO₃ nanoparticles¹² and thin films,¹³ covering x = 0.025 to 0.25. These studies essentially report the enhanced magnetic properties of such films deposited on Si, which are attributed to the ferrimagnetic exchange between the neighboring Fe³⁺ and Ni³⁺ ions.¹² In the present study, we focus on a doping level of 0.5 to 1.5 mol% Ni, corresponding to a range of x = 0.014 to 0.042 in BiFe_1−xNi_xO₃, which pertains to a lower doping level compared with other reported compositions. Furthermore, no report exists yet that discusses surface oxidation and piezoelectric characteristics of the Ni-substituted BFO thin films. Herein, we combine the expected benefit of thin BFO films with the positive effect of Ni doping on the optical and piezoelectric properties. Only 0.5 mol% Ni doping results in a substantial increase in the piezoelectric properties and a reduced band gap. The microstructure and chemical states of the films are examined to explain the influence of the very low Ni content.

2. Experiment

Film preparation

Ni-doped BiFeO₃ thin films were deposited onto a fluorine-doped tin oxide (FTO)–glass substrate by means of spin coating. Bismuth acetate [Bi(CH₃COO)₃ Sigma-Aldrich Inc., USA] and iron acetate [Fe(CH₃COO)₂, Sigma-Aldrich Inc., USA] were dissolved into a solvent containing 2-methoxyethanol (C₃H₈O₂, Sigma-Aldrich Inc., USA) and acetic acid (CH₃COOH, Duksan Chemicals, Korea) to prepare a 0.35 M precursor solution. A 5 mol% excess of Bi was used to compensate for the potential loss of Bi during firing. Ni acetate tetrahydrate [Ni(CH₃COO)₂·4H₂O, Sigma-Aldrich Inc., USA] was dissolved into the BFO solution with differing Ni contents of 0.0, 0.5, 1.0, and 1.5 mol%. Expressed in another way, these compositions correspond to x = 0, 0.014, 0.028, and 0.042 in BiFe_1−xNi_xO₃.

The final precursor solution was spin-coated six times onto a FTO–glass substrate to prepare a ∼200 nm-thick film. After each spin coating, the film was dried at 100 °C for 1 min and then pyrolyzed at 350 °C for 5 min. Finally, the deposited films were annealed at 550–650 °C for 10 min by rapid thermal annealing in an ambient atmosphere.

Characterization

The structure of the annealed films was analyzed by Cu Kα X-ray diffraction (XRD, Max-2500, Rigaku B). The surface morphology was imaged by field-emission scanning electron microscopy (JSM-7001F, JEOL). Transmission electron microscopy (TEM, JEM-ARM-200F, JEOL) combined with energy-dispersive spectroscopy (EDS, X-MaxN 80T, Oxford Instruments) was used to characterize the films at higher magnifications. The valence states of selected elements at the film surface were studied by Al Kα X-ray (1486.6 eV) photoelectron spectroscopy (XPS, K-alpha, Thermo VG). Optical transmission and reflection were measured at room temperature over the spectral range 200–1500 nm by ultraviolet-visible (UV-vis) spectrophotometry (JASCOV 530). The ferroelectric hysteresis loop and piezoelectric characteristics were measured by piezoelectric-force microscopy (PFM, Nanoscope V Multimode, Bruker). Mechanical oscillations were induced by applying an ac field between the tip and holder. We used a Pt/Ir-coated silicon cantilever with a spring constant of ∼42 N m⁻¹ and a resonance frequency of 330 kHz. The driving amplitude and frequency of the ac bias were 5 V rms and 5 kHz, respectively. The local piezoelectric hysteresis loop was measured by applying a dc bias ranging from −10 to +10 V.

3. Results and discussion

Fig. 1(a) shows XRD patterns of undoped BFO thin films annealed at different temperatures. The well-developed crystalline peaks of the rhombohedral BFO phase (JCPDS#: 071-2494) are evident regardless of annealing temperature. The films present no specific growth orientation because of the use of FTO-coated glass substrate. A secondary phase of Bi₂Fe₄O₉ with low peak intensities is observed for annealing temperature at 650 °C, presumably because of the potential volatilization of Bi at high temperatures. The Bi₂Fe₄O₉ phase corresponds to a Bi-deficient phase considering BiFeO₃ stoichiometry. Accordingly, we chose an optimal annealing temperature of 600 °C for subsequent studies.

Fig. 1 XRD patterns of (a) undoped BFO thin films annealed at different temperatures and (b) Ni-doped BFO films annealed at 600 °C, with two insets showing the (012) peak (upon Ni doping) in the 2θ range of 21.8° to 23.2° and of the (104) and (110) peaks in the 2θ range of 31° to 32.7°.

Fig. 1(b) shows the XRD patterns of films annealed at 600 °C and doped with different amounts of Ni (from 0.5 to 1.5 mol%). The patterns are fairly consistent regardless of Ni content, with a strong orientation along the (012) plane. Peak traces of the Bi₂O₃ phase were observed only in the 1.5 mol% Ni sample. Because the Ni-ion substitutes for the Fe site,^12–14 the excess Bi may precipitate out as Bi₂O₃ at higher levels of Ni doping. Depending on the Ni content, the existence of Ni also seems to affect the degree of crystallinity. The intensity variation of the (012) peak in the 2θ range of 21.8–23.2°, as shown in the inset of Fig. 1(b), indicates the significant progress of crystallinity even with a small amount of Ni. The highest-intensity (012) peak occurs with 0.5 mol% Ni. In addition, the (104) and (110) peaks are completely separate in undoped BFO thin film, but the peaks merge gradually into a single peak upon Ni doping. These results imply that with low level Ni doping, the rhombohedral structure is distorted toward a monoclinic or tetragonal structure.⁶

The effect of film thickness on phase evolution is shown in Fig. S1 of the ESI.† Increasing the film thickness from 50 to 200 nm enhances the crystallinity of the 0.5 mol% Ni–BFO film and reveals clear evidence of highly preferred (012) growth.

Selected SEM micrographs showing the surface microstructure of Ni–BFO thin films annealed at 600 °C are shown in Fig. 2(a)–(c). These micrographs show well-grown grain structures with a relatively uniform distribution of grain size that depends on the Ni content. The incorporation of Ni seems to initially increase grain size up to 0.5 mol% and then to decrease it significantly. The largest grain size at 0.5 mol% Ni indicates that there is an optimal amount for promoting grain growth of films. This effect of Ni is very similar to the reported cases of thin films¹³ and bulk ceramics^15,16 containing Ni additive. It seems that Ni acts initially a grain growth promoter via a liquid phase sintering mechanism and then limits grain growth beyond the optimal amount.¹⁵ An excessive amount of Ni may induce a energy barrier for diffusion and grain boundary movement, resulting in less effectiveness in facilitating grain growth.

Fig. 2 Surface SEM images of (a) undoped, (b) 0.5 mol%, and (c) 1.5 mol% Ni-doped BFO thin films and (d) a representative cross-sectional image of 0.5 mol% Ni–BFO thin films on FTO–glass substrate. (e) TEM image of 0.5 mol% Ni–BFO thin film with (f) a diffraction pattern. All samples were annealed at 600 °C.

Fig. 2(d) shows a cross section of a 0.5 mol%, 200 nm-thick Ni–BFO film. Throughout the cross-section, a clear interface is seen between the Ni–BFO and the thick FTO layer. Comparing the large grain sizes of >200 nm with a small thickness of ∼200 nm indicates that the grains are likely to grow in the x–y planar direction because of the thickness constraint. A TEM image in Fig. 2(e) for the 0.5 mol% Ni–BFO film prepared using a standard focused-ion-beam (FIB) technique essentially confirms the dominance of the single-grain structure. The TEM image shows a rough interface between the FTO and BFO layers, which follows the roughness of the FTO-coated substrate itself. The additional high-resolution image presented in Fig. 2(f) shows a selected-area electron-diffraction (SAED) pattern representing rhombohedral BFO. This image reveals the dominant (012) direction of film growth. Note that TEM elemental mapping over the film does not reveal any evidence of Ni segregation at this scale.

To characterize the chemical states at the surface, the valence states of elements of undoped and Ni–BFO samples were examined in detail. High-resolution XPS spectra of the Bi 4f, Fe 2p, Ni 2p, and O 1s regions are shown in Fig. 3(a)–(d), respectively. The XPS spectra in Fig. 3(a) show a doublet of the Bi 4f state that is de-convoluted into the peaks of Bi 4f_7/2 and Bi 4f_5/2 at ∼158.9 and ∼164.3 eV, respectively. The spin-orbit-splitting energy is ∼5.4 eV, which is consistent with the published theoretical value for stoichiometric BFO.^17,18 No evidence appears of Bi deficiency in the samples, suggesting good compensation of the potential Bi loss with the excess of 5 mol% Bi used here. The Fe region shown in Fig. 3(b) consists of two peaks at ∼710.3 and ∼723.9 eV. These peaks come from the Fe 2p_3/2 and Fe 2p_1/2 transitions, respectively, and are attributed to Fe–O bonds. The spin-orbit-splitting energy of the Fe 2p doublet is 13.3 eV, which is consistent with the theoretical value of 13.6 eV for Fe₂O₃.¹⁹ Note that satellite peaks appear 8 eV above the 2p_3/2 peak, indicating the 3+ oxidation state of Fe.^20,21 Fig. 3(c) reveals the Ni 2p-binding-energy region. The peaks of binding energy of Ni 2p_1/2 and 2p_3/2 at ∼872.1 and ∼854.5 eV represent the 2+ ionic state of Ni.¹⁹ The small hump at ∼861.1 eV corresponds to the satellite state of Ni 2p.^22,23 Asymmetric and broad peaks appear in the O 1s spectrum, as shown in Fig. 3(d). These are resolved into three distinct peaks at ∼529.6, 531.4, and 532.8 eV and are attributed to three different O species: O²⁻ ions (O_L) participating in Bi–O bonds in the perovskite, O defects (O_D), and absorbed O (O_a) in weakly bonded O species such as –CO₃ and –OH.²¹ By calculating the ratio of the areas under the two peaks O_D and O_L, we find that the oxygen-vacancy concentration increases with Ni, with values 0.10, 0.17, and 0.21 for undoped, 0.5, and 1.5 mol% Ni–BFO thin films, respectively. This result indicates that Ni substitution intensifies the oxygen-related defects, which is consistent with what is observed in other Ni-doped BFO systems.^12,13,24,25

Fig. 3 XPS spectra of the as-annealed undoped, 0.5 mol% and 1.5 mol% Ni-doped BFO thin films in the binding energy regions of (a) Bi 4f, (b) Fe 2p, (c) Ni 2p, and (d) O 1s peaks.

Fig. 4 shows UV-vis transmittance spectra collected from the Ni–BFO films annealed at 600 °C. The transmittance depends on the Ni content. The 0.5 mol% Ni–BFO thin films have a higher transmittance below a wavelength of ∼1200 nm. This higher transmittance may be due to the increase in grain size with improved densification (less scattering occurs with a lower density of grain boundaries). Plots of (αhν)² versus hν are shown in Fig. 4(b)–(d) for undoped and doped films based on their transmittance and reflectance spectra. Each film's optical band gap E_g was estimated by extrapolating the linear portion of (αhν)² versus photon energy (hν), where α is the absorption coefficient. By using the known values of film thickness d, transmittance T, and reflectance R, α was calculated from the expression α = (1/d)ln[(1 − R)²/T]. This results in E_g = 2.83 eV for the undoped film, which is similar to published values for BFO thin films.^3,4 The optical band gaps of the films tend to decrease with Ni doping: E_g = 2.80 and 2.79 eV are calculated for the samples with 0.5 and 1.0 mol% Ni, respectively. It is plausible that the high concentration of Ni increases the concentration of oxygen-related defect states in the valence-band edge, thereby widening the valence band and thus contributing to narrowing the band gap,^24,26 as suggested by the XPS result shown in Fig. 3(d). The effect of film thickness on UV-vis spectra can be seen in Fig. S2 of the ESI.† As expected, a higher transmittance (∼95%) occurs for the thinner films of 50 nm at 700 nm. A higher optical band gap of 3.07 eV is obtained for the 0.5 mol% 50 nm-thick Ni–BFO film.

Fig. 4 (a) UV-visible transmittance spectra, and the (αhν)² vs. hν curves of (b) 0.0, (c) 0.5, and (d) 1.0 Ni-doped BFO films annealed at 600 °C.

Fig. 5 shows the piezoelectric response (PR) and phase curves as a function of dc bias over the range from −10 V to +10 V for undoped and Ni-doped samples, which were measured by PFM. The PR amplitude loops resulting from the applied voltages appear as butterfly-shaped hysteresis curves for the undoped and doped thin films, corresponding to the strain–electric-field (S–E) behavior of piezoelectric materials [Fig. 5(a)–(c)]. This indicates the nonlinear nature of BFO thin films with the relationship between the Ni content and the piezoelectricity. However, the phase curves of the films do not reveal significant differences [Fig. 5(d)–(f)]. The difference in PR phase (ΔΦ) with the opposite signal is about 150° in these cases, which implies that a 150° domain switch occurs upon applying the dc bias. The applied voltage of 10 V corresponds to an electric field of ∼50 kV mm⁻¹ for a film thickness of ∼200 nm. The range of electric field is in the reported range of the saturated field (33–100 kV mm⁻¹) for BFO thin films.^27,28 The coercive voltage V_c was evaluated by using the relation V_c = (V_c⁺ − V_c⁻)/2, where V_c⁺ and V_c⁻ are the starting points for switching polarization by applying positive and negative bias, respectively (see derivation elsewhere²⁹). Table 1 gives V_c for Ni-doped films normalized by that of undoped Ni sample. The result for V_c seems to increase with Ni doping. It is plausible that Ni²⁺ substitutes into the B site and induces the hard doping effect that results in the increase in coercive field, similar to what occurs in piezoelectric thin films.^30,31 These results indicate that a stronger electric field is required for domain reversal with Ni doping.

Fig. 5 Variation in piezoelectric properties of BFO thin films with Ni doping, as represented by (a)–(c) piezoresponse and (d)–(f) phase curves with dc bias for undoped, 0.5 mol%, and 1.5 mol% Ni-doped BFO thin films, respectively.

Table 1 Normalized coercive field and piezoelectric coefficient of Ni-doped BFO thin films, as obtained by piezoelectric-force microscopy

Ni (mol%)	0.0	0.5	1.5
Normalized Vc	1.00	1.18 ± 0.2	1.05 ± 0.1
d33,eff (pm V^−1)	15.4 ± 0.91	28.0 ± 1.9	26.9 ± 1.6

---

Meanwhile, the magnitude of the PR was considerably enhanced with higher Ni doping. The piezoelectric coefficient d_33,eff was calculated by using the relation Δu = δA = d_33,effV_ac, where Δu is the piezoresponse displacement, A is the resulting PR amplitude, δ is the deflection sensitivity of the cantilever, and V_ac is the amplitude of the driving ac voltage.³² The average values are also listed in Table 1. Larger piezoelectric coefficients occur upon adding Ni. The coefficient reaches 28.0 pm V⁻¹ at 0.5 mol% of Ni, which is about twice that for the undoped sample. This significant improvement may be attributed to the enhanced crystallinity and densification upon Ni doping. The increased crystallinity may reduce the density of grain boundaries so that the movement of domain walls is less restricted by grain boundaries.^33,34 Another possible explanation is a Ni-doping-driven structural phase transition in the films, as observed in the XRD results. The piezoelectric constant of BFO is known to increase substantially at the structural boundary: the morphotropic-phase region.³⁵ The structural change and increased crystallinity of BFO thin films upon Ni doping thus change the piezoelectric properties.

4. Conclusions

Doping with Ni has been found to be desirable because it reduces the optical band gap and increases the piezoelectric properties for highly transparent BiFeO₃ films. The d₃₃ coefficient increases about twofold upon Ni doping. The achievements are surprising because they occur for just a very small amount of doping (0.5 mol% Ni). The doping was found to induce a substantial increase in crystallinity and thus reduce the negative effects of grain boundaries. Analyses by XPS reveal that Ni doping generates more oxygen-related defects, which are responsible for the reduced optical band gap.

Acknowledgements

This work was supported by a grant from the National Research Foundation of Korea (NRF-2011-0020285 and NRF-2013R1A2A2A01016711).

References

1. S. Gupta, M. Tomar and V. Gupta, J. Appl. Phys., 2014, 115, 014102.
2. W. Sun, J.-F. Li, Q. Yu and L.-Q. Cheng, J. Mater. Chem. C, 2015, 3, 2115–2122.
3. X. Chen, H. Zhang, T. Wang, F. Wang and W. Shi, Phys. Status Solidi A, 2012, 209, 1456–1460.
4. W. Ji, K. Yao and Y. C. Liang, Adv. Mater., 2010, 22, 1763–1766.
5. F. Wang, I. Grinberg and A. M. Rappe, Appl. Phys. Lett., 2014, 104, 152903.
6. P. Kharel, S. Talebi, B. Ramachandran, A. Dixit, V. M. Naik, M. B. Sahana, C. Sudakar, R. Naik, M. S. R. Rao and G. Lawes, J. Phys.: Condens. Matter, 2009, 21, 036001.
7. L. Kang, W. Zhang, Y. Sun and J. Ouyang, Phys. Status Solidi A, 2014, 211, 565–569.
8. F. Zavaliche, R. R. Das, D. M. Kim, C. B. Eom, S. Y. Yang, P. Shafer and R. Ramesh, Appl. Phys. Lett., 2005, 87, 182912.
9. Y. Guo, B. Guo, W. Dong, H. Li and H. Liu, Nanotechnology, 2013, 24, 275201.
10. W. Dong, Y. Guo, B. Guo, H. Li, H. Liu and T. W. Joel, ACS Appl. Mater. Interfaces, 2013, 5, 6925–6929.
11. W. Dong, Y. Guo, B. Guo, H. Liu, H. Li and H. Liu, Mater. Lett., 2012, 88, 140–142.
12. J. Zhao, X. Zhang, S. Liu, W. Zhang and Z. Liu, J. Alloys Compd., 2013, 557, 120–123.
13. S. K. Singh, R. Palai, K. Maruyama and H. Ishiwara, Electrochem. Solid-State Lett., 2008, 11, G30–G32.
14. Q. Xu, M. Sobhan, F. Anariba, J. W. C. Ho, Z. Chen and P. Wu, Phys. Chem. Chem. Phys., 2014, 16, 23089.
15. Y. C. Huang and W. H. Tuan, Mater. Chem. Phys., 2007, 105, 320–324.
16. W. Cai, J. Gao, C. Fu and L. Tang, J. Alloys Compd., 2009, 487, 668–674.
17. Z. Quan, H. Hu, S. Xu, W. Liu, G. Fang, M. Li and X. Zhao, J. Sol-Gel Sci. Technol., 2008, 48, 261–266.
18. X. Ou, Y. Shuai, W. Luo, P. F. Siles, R. Kogler, J. Fiedler, H. Reuther, S. Zhou, R. Hubner, S. Facsko, M. Helm, T. Mikolajick, O. G. Schmidt and H. Schmidt, ACS Appl. Mater. Interfaces, 2013, 5, 12764–12771.
19. F. Moulder, W. F. Stickle, P. E. Sobol and K. D. Bomben, Handbook of X-Ray Photoelectron Spectroscopy, Perkin-Elmer, Minnesota, 1992.
20. P. S. V. Mocherla, C. Karthik, R. Ubic, M. S. Ramachandra Rao and C. Sudakar, Appl. Phys. Lett., 2013, 103, 022910.
21. 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.
22. L. Zhang and H. Gong, J. Mater. Chem. A, 2015, 3, 7607–7615.
23. K. Srinivas, S. Manjunath Rao and P. Venugopal Reddy, Nanoscale, 2011, 3, 642–653.
24. X. Xu, W. Liu, H. Zhang, M. Guo, P. Wu, S. Wang, J. Gao and G. Rao, J. Appl. Phys., 2015, 117, 174106.
25. Y. J. Yoo, J. S. Hwang, Y. P. Lee, J. S. Park, J.-H. Kang, J. Kim and B. W. Lee, J. Appl. Phys., 2013, 114, 163902.
26. J. Wang, Z. Wang, B. Huang, Y. Ma, Y. Liu, X. Qin, X. Zhang and Y. Dai, ACS Appl. Mater. Interfaces, 2012, 4, 4024–4030.
27. Q. Zhang, N. Valanoor and O. Standard, J. Mater. Chem. C, 2015, 3, 582–595.
28. X. Qi, P.-C. Tsai, Y.-C. Chen, C.-H. Ko, J.-C.-A. Huang and I. –G. Chen, J. Phys. D: Appl. Phys., 2008, 41, 232001.
29. H. B. Kang, J. Chang, K. Koh, L. Lin and Y. S. Cho, ACS Appl. Mater. Interfaces, 2014, 6, 10576–10582.
30. H. E. Mgbemere, M. Hinterstein and G. A. Schneider, J. Eur. Ceram. Soc., 2012, 32, 4341–4352.
31. W. Zhu, I. Fujii, W. Ren and S. Trolier-Mckinstry, J. Appl. Phys., 2011, 109, 064105.
32. R. García and R. Pérez, Surf. Sci. Rep., 2002, 47, 197–301.
33. W. Cao and C. A. Randall, J. Phys. Chem. Solids, 1996, 57, 1499–1505.
34. H. T. Martirena and J. C. Burfoot, J. Phys. C: Solid State Phys., 1974, 7, 3182–3192.
35. D. Kan, C. J. Long, C. Steinmetz, S. E. Lofland and I. Takeuchi, J. Mater. Res., 2012, 27, 2691–2704.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra27674h

---