Transition of resistive switching behavior for a HfO_x film induced by Ni doping

Abstract

In this work, Ni-doped HfO_x films with different Ni doping concentrations were fabricated and the chemical bonding states as well as the resisistive switching characteristics were investigated. The high Ni concentration of 5.3% doping into a HfO_x sample showed the improved switching performance including enlarged ON/OFF ratio and reduced switching voltages. The switching behavior transformed from bipolar to unipolar resisistive switching due to the higher Ni doping concentration of 8.4%. The bipolar resisistive switching behavior can be attributed to the formation and recovery of oxygen vacancy filaments in the film, while the unipolar resisistive switching behavior was due to the excessive Ni dopants in the film. The C-AFM measurement was performed to observe the evolution of conductive filaments and a physical model based on oxygen vacancies or metal filaments was constructed. The reliability properties for all the samples were also demonstrated.

---

Introduction

Traditional flash memories have been reaching their scaling limits, for example, it becomes increasingly difficult to maintain a sufficient number of electrons in these charge storage-based memories with the coming of a technology node of integrated circuit. Many novel non-volatile memories have been proposed as an alternative.^1–3 Recently, resistance random access memory (RRAM) has attracted significant attention for its features such as simple structure, low power consumption, and good compatibility with complementary metal oxide semiconductor (CMOS) processes.^4,5 Based on the dependence on the polarity of operation voltage, resistive switching (RS) can be classified into two types: bipolar resistive switching (BRS) and unipolar resistive switching (URS).^4,6 The BRS behavior depends on the polarity of applied voltage, while the URS shows the opposite behavior which depends on the amplitude of the applied voltage but not on the polarity.⁷ As a kind of binary transition metal oxides, hafnium oxide (HfO₂) not only has wide applications in optoelectronics and high-k dielectric materials,^8,9 but also has been investigated intensively in RRAM due to the its well-known fabrication technology and good thermodynamic stability.¹⁰ At present, large memory window and retention property have been reported for HfO₂-based RRAM.^11,12 However, one major challenge for its practical application in RRAM is how to the minimize the variations of switching parameters.¹³ Since the switching behavior has been widely identified to be closely related with the formation and rupture of oxygen vacancies- or metal ions-based conductive filaments,^14,15 one can expect that the precise control of these defects can effectively improve the device performance. The incorporation of impurities with a different bonding valence into oxide film would be more effective in changing the defects in the film.¹⁶ The RS properties in metal doped HfO₂ film have been reported previously.^17,18 However, the switching characteristics of Ni-doped HfO_x films and the role of Ni dopants are rarely studied. In this work, the Ni-doped HfO_x films were fabricated by radio frequency magnetron sputtering and the effects of Ni doping concentration on chemical composition and RS behaviors of HfO_x films were systematically investigated. The transition of RS behaviors induced by the higher Ni doping concentration was demonstrated. The switching mechanisms for different RS behaviors were discussed based on conductive filament model and the evolution of filaments was also explored.

Experimental methods

The Si/SiO₂/Ti/Pt substrates with size of 10 mm × 10 mm were used as the bottom electrodes and the thicknesses for Si/SiO₂, Ti and Pt were 300 nm, 30–50 nm and 150 nm respectively. The Pt substrates were cleaned by deionized water, alcohol and acetone successively before deposition. Then the amorphous HfO_x and Ni-doped HfO_x films with size of 8 mm × 8 mm were deposited on Pt substrates by radio frequency magnetron sputtering and the film thicknesses were 20 nm. The different types of Ni-doped HfO_x films were fabricated by co-sputtering the metal Hf and Ni targets with O₂ as the reactive gas, and the Ni doping concentration was controlled by modulating the area of Ni target. The working pressure and the sputtering power during deposition were 0.3 Pa and 80 W respectively. The rapid thermal annealing process was performed in N₂ atmosphere for 10 min with the temperature of 200 °C. The chemical bonding states of the films were analyzed by X-ray photoelectron spectroscopy (XPS), and the Ni doping concentrations were estimated to be 0%, 2.4%, 5.3% and 8.4% respectively. Finally, to measure the electrical properties of the films, the Cu top electrodes with diameter of 2 mm were deposited by evaporation with a metal shadow mask. The electrode thickness was 50 nm. The electrical properties were measured using a two-probe method with an Agilent 4155C semiconductor parameter analyzer.

Results and discussion

To investigate the chemical composition of the prepared HfO_x films, the XPS analyses were carried out. All peaks are calibrated by C 1s peak (284.6 eV). Fig. 1a shows the core level spectra of Hf 4f in HfO_x films with different Ni doping concentrations. The Hf 4f spectra shift slightly to the higher binding energies as the Ni doping concentration increases, which correspond to Hf–O. It can be assumed that doping Ni into the HfO_x film does not cause large chemical reaction. The spectra of Ni 2p in Ni-doped HfO_x films are presented in Fig. 1b. For 2.4% Ni-doped HfO_x film, two peaks at 855.3 eV and 857.8 eV can be observed, which correspond to the Ni²⁺ and Ni³⁺ bonding states respectively.^19,20 The increase of Ni doping concentration results in the decrease of intensity for Ni³⁺ peak. As the Ni doping concentration reaches a higher value of 8.4%, a peak at 852.8 eV which corresponds to the metallic Ni^16,19 is detected. The weak shift of Hf 4f peak may be caused by the substitution of Ni atoms on Hf sites, which induces new oxygen vacancies in HfO_x film.²¹ Besides, parts of Ni dopants occupy the interstitial sites, compensating the defects in the film.

Fig. 1 The XPS spectra of (a) Hf 4f and (b) Ni 2p of Ni-doped HfO_x films.

The RS behaviors of HfO_x samples with different Ni doping concentrations are presented in Fig. 2. The inset of Fig. 2a shows the schematic diagram of Cu/HfO_x/Pt (MIM) structures for I–V measurements. The bias voltage was applied on Cu top electrode and the Pt bottom electrode was grounded. The initial resistance state of the prepared samples were all in high resistance state (HRS). A forming process was required to activate the RS behavior, as shown in Fig. 2a. With the increase of Ni doping concentration, the forming voltage decreased gradually, which may be due to the increased oxygen vacancies induced by Ni doping. Note that the initial currents for 8.4% Ni-doped HfO_x sample increased, which was related to the existence of metal Ni in the film. After the forming process, the sample can reversibly switch between HRS and low resistance state (LRS). The current compliance of 10 mA was applied during the set process to prevent the sample from damage and the bias voltage swept in a counterclockwise direction. The typical BRS behavior was observed for the 2.4% and 5.3% Ni-doped HfO_x samples, as presented in Fig. 2b and c respectively. The ON/OFF ratio for 2.4% Ni-doped HfO_x sample decreased after dozens of cycles. While, 5.3% Ni-doped HfO_x sample exhibited good reproducibility with ON/OFF ratio close to 10³. With the increase of Ni doping concentration, an interesting finding was achieved. The 8.4% Ni-doped HfO_x sample can exhibit both BRS and URS behaviors without forming process. However, the applied positive bias on the device led to the breakdown of sample after several switching cycles. The URS behavior for 8.4% Ni-doped HfO_x sample is presented in Fig. 2d. The switching voltages were both in negative direction. It has been reported that some oxide systems with Ni as one of the electrodes shows URS behavior due to the diffusion of Ni ions into the oxide film.²² In this work, the transition of switching behavior from BRS to URS due to the higher Ni doping concentration must be related to the excessive Ni dopants in HfO_x film.

Fig. 2 (a) The forming process of the undoped and Ni-doped HfO_x samples. The typical I–V curves of HfO_x samples with different Ni doping concentrations of (b) 2.4%, (c) 5.3% and (d) 8.4% respectively. The inset of (a) shows the schematic diagram for I–V measurements.

Fig. 3a shows the Weibull plots of switching voltages of the undoped and Ni-doped HfO_x samples. The gradually decreased switching voltages were observed for the Ni-doped HfO_x samples with BRS behavior. For 8.4% Ni-doped HfO_x sample, although the switching voltages were large, no overlap of V_set and V_reset can be observed. For the device with URS behavior, one of the crucial issues is to guarantee the non-overlap of V_set and V_reset. The Weibull plots of the resistances in HRS and LRS for the prepared HfO_x samples are presented in Fig. 3b. The narrower distribution of resistances for Ni-doped HfO_x samples can be achieved. Besides, the resistances in HRS for Ni-doped HfO_x samples increased, which resulted in the enlarged ON/OFF ratio. Based on the XPS and electrical analysis, the substitution of Hf by Ni atoms induced oxygen vacancies in the film which decreased the switching voltages and suppressed the random distribution of oxygen vacancies. While parts of Ni atoms which incorporated as an interstitial site could be scattering centers, increasing the resistivity of the film.²³ Thus, the resistance in HRS increased with Ni doping concentration (<5.3%). However, the higher Ni doping concentration caused large numbers of defects and the metal Ni in the film. As a result, the resistance in HRS for 8.4% Ni-doped HfO_x sample decreased.

Fig. 3 The Weibull plots of (a) switching voltages and (b) resistances in HRS and LRS of the undoped and Ni-doped HfO_x samples.

To explore the origin of switching behaviors of HfO_x samples with different Ni doping concentrations. The I–V curves are replotted in double-log scale, as shown in Fig. 4a. The inset of Fig. 4a shows the current conduction of LRS for the prepared HfO_x samples. A linear relationship between current and voltage can be fitted, indicating that the currents in LRS are controlled by ohmic conduction. The I–V curves for HRS are a little complicated, as shown in Fig. 4, and can be fitted as ohmic region at low voltage and Child's law region at higher voltage, which can be well explained by the space charge limited current (SCLC) effect.^14,24 Metal ions and oxygen vacancies are reported to play important role on the current conduction in the film. In our work, the migration of oxygen vacancies or metal ions (Cu, Ni) under set voltage forms conductive filaments in HfO_x film and the sample transforms from HRS to LRS. By applying the reset voltage, the filaments ruptured and the sample switches back to HRS. To explore the composition of conductive filaments, the dependence of resistance on temperature is measured, as shown in Fig. 4b. For the samples with Ni doping concentration lower than 5.3% (BRS), the negative temperature dependence of resistance in LRS indicates the semiconducting behavior. While for 8.4% Ni-doped HfO_x sample, as shown in the inset of Fig. 4b, the resistance in LRS increases with the increase of temperature, exhibiting the metallic characteristic. The resistance in HRS for all samples decreases as the temperature increases. It can be deduced that the compositions of conductive filaments for BRS and URS are mainly oxygen vacancies which assist the charge hopping²⁵ and metal ions respectively. In addition, an obvious difference of reset process can be observed for BRS and URS with gradual and abrupt switching respectively. For BRS, the gradual switching suggests a gradual rapture of the filament, indicating the recovery of oxygen vacancy chain during reset process. While the abrupt reset process for URS would be attributed the rupture of a metal filament by Joule heat.

Fig. 4 (a) The switching mechanisms and (b) the dependence of resistances on temperature for the undoped and Ni-doped HfO_x samples. (c) The C-AFM current maps of 5.3% Ni-doped HfO_x sample. The inset of (a) shows the fitting curves of LRS. The inset of (b) shows the variation of resistance for 8.4% Ni-doped HfO_x sample.

In order to further investigate the switching behavior, the conductive atomic force microscopy (C-AFM) measurements were performed. The MIM structure for CAFM measurement was similar to, but not exactly the same with, I–V measurement. For I–V measurement, the MIM structure was Pt/HfO_x:Ni/Cu, whereas for CAFM measurement, the MIM structure was Pt/Cu/HfO_x:Ni/CAFM-tip (Pt). Fig. 5c presents the C-AFM image of 5.3% Ni-doped HfO_x sample. For initial sample, no obvious current can be detected in C-AFM current map at a low bias voltage <0.5 V. By applying the voltage of 2 V, the bright regions occurred in set sample, which represented the localized high conductive regions, indicating the formation of conductive filaments and the switching of sample from HRS to LRS.²⁶ By applying the reset voltage, the bright regions vanished, indicating the rupture of filaments. Note that after the reset process, the sample did not return to its initial state. There still existed some tiny conductive filaments, and that is why the current for HRS is lager than that of initial state.

Fig. 5 The schematic diagrams of switching mechanism for (a) BRS and (b) URS behaviors.

The physical models for different switching processes are proposed. Fig. 5a shows the schematic diagram of switching mechanism for BRS. The formation and recovery of oxygen vacancies filaments were dominantly responsible for the BRS behaviors. During the set process, the migration of oxygen vacancies to cathode under the positive voltage resulted in the formation of oxygen vacancies filaments, switching the sample into LRS. By applying the reverse voltage, the oxygen vacancies recovered and the sample was switched to HRS again. This process is possibly accompanied with the redox of metal ions with the increase of Ni doping concentration. The RS mechanism for URS is presented in Fig. 5b. The dominant role of conductive channels can be attributed to the formation of filaments by the migration of metal ions (Ni, Cu) between two metal electrodes and the rupture of filaments by Joule heating. By applying the set voltage, the metal ions moved towards cathode and were reduced to metal. The accumulation of Ni or Cu metal towards bottom electrode led to the formation of metal filaments in the film. During the reset process, as the current increased, a large number of joule heating was generated to increase the temperature of the film,²⁷ which led to the rupture of the filaments near interface. Then, the sample was switched back to HRS. The transition of RS behavior is still under debate and a further understanding of the effect of the impurity doping on RS behavior is required.

The reliability of HfO_x samples with different Ni doping concentrations were investigated. The endurance properties are presented in Fig. 6a. The bigger fluctuations of currents in HRS and LRS were observed for HfO_x sample (0% Ni), which was due to the random distribution of oxygen vacancies in the film. The Ni-doped HfO_x samples revealed more stable switching behavior over 100 cycles. The least ON/OFF ratio can be up to 10² which was large enough to distinguish the two resistance states. Fig. 6b indicates the retention behaviors of the prepared samples at room temperature. The two states can maintain for 10⁴ s without degeneration for all samples. The good RS performance indicates the possibility of Ni-doped HfO₂ samples for the non-volatile memory application.

Fig. 6 (a) The endurance and (b) the retention properties of HfO_x samples with different Ni doping concentrations. The tests were performed at room temperature.

Conclusions

In summary, the chemical bonding states and the RS behaviors of Ni-doped HfO_x sample with different Ni doping concentrations were investigated. The HfO_x film with high Ni doping concentration (5.3%) exhibited improved BRS behaviors including enlarged ON/OFF ratio, decreased switching voltages and good endurance and retention properties. The switching behavior transformed from BRS to URS when the Ni doping concentration reached a higher value (8.4%). The switching mechanism for BRS and URS behaviors can be ascribed to the different conductive filaments. The BRS behavior was attributed to the formation and recovery of oxygen vacancy filaments in the film, while the URS behavior was related to the excessive Ni dopants in the film and the filaments were disrupted by Joule heat.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (No. 51202196), the Fundamental Research Funds for the Central Universities (No. 3102014JCQ01032), the 111 Project (No. B08040), and the Innovation Foundation for Doctor Dissertation of Northwestern Polytechnical University (No. CX201612).

References

1. D.-H. Lim, G.-Y. Kim, J.-H. Song, K.-S. Jeong, D.-C. Kim, S.-W. Nam, M.-H. Cho and T.-G. Lee, RSC Adv., 2015, 5, 221–230.
2. J. K. Lee, S. Jung, J. Park, S.-W. Chung, J. S. Roh, S.-J. Hong, l. H. Cho, H.-I. Kwon, C. H. Park, B.-G. Park and J.-H. Lee, Appl. Phys. Lett., 2012, 101, 103506.
3. Y. Wang, Q. Liu, S. Long, W. Wang, Q. Wang, M. Zhang, S. Zhang, Y. Li, Q. Zuo, J. Yang and M. Liu, Nanotechnology, 2010, 21, 045202.
4. R. Waser and M. Aono, Nat. Mater., 2007, 6, 833–840.
5. L. Zhang, Y.-Y. Hsu, F. T. Chen, H.-Y. Lee, Y.-S. Chen, W.-S. Chen, P.-Y. Gu, W.-H. Liu, S.-M. Wang, C.-H. Tsai, R. Huang and M.-J. Tsai, Nanotechnology, 2011, 22, 254016.
6. D. Xu, Y. Xiong, M. Tang and B. Zeng, J. Alloys Compd., 2014, 584, 269–272.
7. A. Sawa, Mater. Today, 2008, 11, 28–36.
8. J. M. Khoshmana and M. E. Kordesch, Surf. Coat. Technol., 2006, 201, 3530–3535.
9. C. Mannequin, P. Gonon, C. Vallée, L. Latu-Romain, A. Bsiesy, H. Grampeix, A. Salaün and V. Jousseaume, J. Appl. Phys., 2012, 112, 074103.
10. R. Zazpe, M. Ungureanu, F. Golmar, P. Stoliar, R. Llopis, F. Casanova, D. F. Pickup, C. Rogero and L. E. Hueso, J. Mater. Chem. C, 2014, 2, 3204–3211.
11. K.-L. Lin, T.-H. Hou, J. Shieh, J.-H. Lin, C.-T. Chou and Y.-J. Lee, J. Appl. Phys., 2011, 109, 084104.
12. H. Zhang, L. Liu, B. Gao, Y. Qiu, X. Liu, J. Lu, R. Han, J. Kang and B. Yu, Appl. Phys. Lett., 2011, 98, 042105.
13. G. I. Meijer, Science, 2008, 319, 1625–1626.
14. F. Yang, M. Wei and H. Deng, J. Appl. Phys., 2013, 114, 134502.
15. H. Zhang, B. Gao, B. Sun, G. Chen, L. Zeng, L. Liu, X. Liu, J. Lu, R. Han, J. Kang and B. Yu, Appl. Phys. Lett., 2010, 96, 123502.
16. J. Kim, H. Na, S. Lee, Y. H. Sung, J. H. Yoo, D. S. Lee, D. H. Ko and H. Sohn, Curr. Appl. Phys., 2011, 11, e70–e74.
17. Y.-S. Fan, P.-T. Liu, L.-F. Teng and C.-H. Hsu, Appl. Phys. Lett., 2012, 101, 052901.
18. W. Lee, J. Park, S. Kim, J. Woo, J. Shin, D. Lee, E. Cha and H. Hwang, Appl. Phys. Lett., 2012, 100, 142106.
19. M. J. Lee, Y. Park, S. E. Ahn, B. S. Kang, C. B. Lee, K. H. Kim, W. X. Xianyu and I. K. Yoo, J. Appl. Phys., 2008, 103, 013706.
20. G. Ma, X. Tang, Z. Zhong, H. Zhang and H. Su, Microelectron. Eng., 2013, 108, 8–10.
21. S. Q. Wang and J. W. Mayer, J. Appl. Phys., 1988, 64, 4711–4716.
22. T.-H. Hou, K.-L. Lin, J. Shieh, J.-H. Lin, C.-T. Chou and Y.-J. Lee, Appl. Phys. Lett., 2011, 98, 103511.
23. H. Li, Q. Chen, X. Chen, Q. Mao, J. Xi and Z. Ji, Thin Solid Films, 2013, 537, 279–284.
24. S. Jou and C.-L. Chao, Surf. Coat. Technol., 2012, 231, 311–315.
25. S. Yu, X. Guan and H. S. P. Wong, Appl. Phys. Lett., 2011, 99, 063507.
26. X. Zhu, W. Su, Y. Liu, B. Hu, L. Pan, W. Lu, J. Zhang and R.-W. Li, Adv. Mater., 2012, 24, 3941–3946.
27. U. Russo, D. Ielmini, C. Cagli, A. Lacaita, S. Spicga, C. Wiemer, M. Perego and M. Fanciulli, IEEE Int. Electron Devices Meet., Tech. Dig., 50th, 2007, 775–778.

---

Footnote

† These authors contributed equally to this work.

---