Bipolar Resistive Switching Characteristics of HfO2/TiO2/HfO2 Trilayer-Structure RRAM Devices on Pt and TiN-Coated Substrates Fabricated by Atomic Layer Deposition
- 927 Downloads
The HfO2/TiO2/HfO2 trilayer-structure resistive random access memory (RRAM) devices have been fabricated on Pt- and TiN-coated Si substrates with Pt top electrodes by atomic layer deposition (ALD). The effect of the bottom electrodes of Pt and TiN on the resistive switching properties of trilayer-structure units has been investigated. Both Pt/HfO2/TiO2/HfO2/Pt and Pt/HfO2/TiO2/HfO2/TiN exhibit typical bipolar resistive switching behavior. The dominant conduction mechanisms in low and high resistance states (LRS and HRS) of both memory cells are Ohmic behavior and space-charge-limited current, respectively. It is found that the bottom electrodes of Pt and TiN have great influence on the electroforming polarity preference, ratio of high and low resistance, and dispersion of the operating voltages of trilayer-structure memory cells. Compared to using symmetric Pt top/bottom electrodes, the RRAM cells using asymmetric Pt top/TiN bottom electrodes show smaller negative forming voltage of −3.7 V, relatively narrow distribution of the set/reset voltages and lower ratio of high and low resistances of 102. The electrode-dependent electroforming polarity can be interpreted by considering electrodes’ chemical activity with oxygen, the related reactions at anode, and the nonuniform distribution of oxygen vacancy concentration in trilayer-structure of HfO2/TiO2/HfO2 on Pt- and TiN-coated Si. Moreover, for Pt/HfO2/TiO2/HfO2/TiN devices, the TiN electrode as oxygen reservoir plays an important role in reducing forming voltage and improving uniformity of resistive switching parameters.
KeywordsAtomic layer deposition Resistive random access memory Bottom electrode Resistive switching parameters Oxygen vacancy concentration Trilayer structure
Resistive random access memory (RRAM) has attracted great attention due to its potential for the replacement of flash memory in next-generation nonvolatile memories [1, 2, 3]. Resistive switching (RS) phenomenon has been widely discovered in transition metal oxides, solid electrolytes, and organic polymers [4, 5, 6, 7]. RRAM devices based on transition metal oxides have been extensively explored because of its simple composition and compatible processing with conventional complementary metal-oxide-semiconductor (CMOS) fabrication [8, 9, 10]. The filament model of oxygen vacancy migration is used to elucidate the switching behavior [1, 11]. A unified microscopic principle is proposed to quantify both unipolar and bipolar switching characteristics of transition metal oxide-based RRAM, which are correlated with the distribution of localized oxygen vacancies in the oxide switching layer [12, 13].
However, memory cells using transition metal oxides suffer from nonuniformity of resistive switching parameters, such as unstable resistance values of low and high resistance states (LRS and HRS), dispersed set and reset voltages, impeding the commercial applications. Recently, the trilayer-structure oxide-based RRAM devices have been demonstrated to improve the dispersion of resistive switching parameters. The cells with a structure of Al2O3/HfO2/Al2O3 exhibited fantastic uniformity of set and reset voltages, and excellent endurance of switching between the LRS and HRS . The linkage or rupture of the conductive filaments more easily occurred in two interfacial layers between Al2O3/IL/HfO2/IL/Al2O3. Meanwhile, the unit of trilayer-structure of TaOx/TiO2/TaOx showed good performance in one selector-one resistor arrays, which was ascribed to the fact that the energy band of the TiO2 film was symmetrically bent at the top and bottom TaO x /TiO2 interfaces and modified as a crested oxide barrier due to some Ta atoms diffusion into TiO2 film .
Moreover, the RS behavior of a given oxide storage medium can be significantly affected by the electrode materials [1, 16, 17]. However, the existing models based on the free energy of interfacial oxide formation and the metal work functions are insufficient to completely explain the results. Meanwhile, the work on the electrodes dependent RS phenomenon of the trilayer-structure RRAM is also rather lacking at present.
Atomic layer deposition (ALD) is a new type of thin film deposition technology based on sequential self-limited and complementary surface chemisorptions reactions using precursor vapor with simple and precision thickness control, large area uniformity, and excellent three-dimensional conformality, especially for deposition of nano-laminated structure [18, 19].
In this work, the HfO2/TiO2/HfO2 trilayer-structure RRAM devices have been prepared on Si/SiO2/Ti/Pt and Si/TiN substrates with Pt top electrodes by ALD. The impact of the bottom electrodes of Pt and TiN on the RS behaviors of HfO2/TiO2/HfO2 devices has been investigated carefully. The related explanation has been proposed.
In this experiment, we used two different bottom electrodes, including commercial Si/SiO2/Ti/Pt and home-made Si/SiO2/TiN. Conductive TiN was deposited by plasma-enhanced atomic layer deposition (PEALD) in our laboratory.
ALD was performed in a commercial Picosun SUNALETMR-200 advanced reactor (Picosun, Finland). P type Si (100) wafers with a resistivity of 1~10 Ω cm were used as the starting substrates. After the conventional RCA cleaning of the Si wafers without removing native oxide, 30-nm-thick TiN was deposited on Si as the bottom electrode at 400 °C by PEALD, using room temperature TiCl4 and NH3 plasma gas as the Ti and N precursors, respectively. Liquid NH3 was selected as NH3 plasma source at room temperature. The plasma power and NH3 gas flow rate were 2500 W and 150 sccm, respectively.
Subsequently, 5 nm HfO2/10 nm TiO2/5 nm HfO2 stacking structures were deposited in turn on Pt- and TiN-coated Si substrates at 250 °C by thermal ALD using Hf[N(C2H5)CH3]4 (TEMAH), TiCl4, and H2O as the Hf, Ti, and O precursors, respectively, where one oxide cycle consisted of 0.1s metal source injection, 4s N2 purging, 0.1s H2O injection, and 4s N2 purging. TEMAH was evaporated at 150 °C. Pure N2 (99.999%) was used as carrier gas and purging gas. Then, 100-nm-thick Pt top electrodes were DC sputtered through a shadow mask with a diameter of 150 μm using the Q150T system.
The growth per cycle (GPC) of pure HfO2 or TiO2 on Si was determined by spectroscopic ellipsometer (GES-5, Sopra). The topography and surface roughness of the films and bottom electrodes were analyzed by atomic force microscopy (AFM, Cypher, Asylum Research). The root-mean-square (RMS) roughness values were recorded from 1 μm × 1 μm areas. The composition and chemical state of the stack structures were examined by X-ray photoelectron spectroscopy (XPS, Thermo Fisher K-Alpha) with a monochromatic Al Kα source (hν = 1486.6 eV) for excitation of photoelectrons. The charge effect was calibrated by setting the C 1s photoemission at 284.6 eV. The XPS depth profile of HfO2/TiO2/HfO2 on Pt- and TiN-coated Si was obtained by Ar ion etching. The electrical properties of the HfO2/TiO2/HfO2 trilayer-structure RRAM devices were measured by Keithely 4200 semiconductor characterization system on probe station (CasCade Summit 12000 B-M). A current compliance of 10 mA was imposed to protect the fabricated device units from damages of high current during set processes. The bias voltage was applied to the Pt top electrode with the grounded bottom electrodes of Pt or TiN.
Results and Discussion
The trilayer-structure RRAM devices with symmetric Pt top/bottom electrodes and asymmetric TiN bottom electrode/Pt top electrodes exhibit such different electroforming polarity preference. The bottom electrode of chemically inert Pt or relatively active TiN seems to play a key role. The related reason on electrode-dependent electroforming polarity will be discussed later after considering the XPS depth profiles of trilayer structures of HfO2/TiO2/HfO2 on Pt- and TiN-coated Si.
For Pt/HfO2/TiO2/HfO2/Pt and Pt/HfO2/TiO2/HfO2/TiN, the resistance level of the ON and OFF states has no evident changes after cumulative waiting time of 104 s at room temperature in Fig. 4c–d, indicating a better retention property of both memory cells over 10 years based on the extrapolation method.
Pt/HfO2/TiO2/HfO2/TiN devices have much lower resistance ratio of OFF/ON states than Pt/HfO2/TiO2/HfO2/Pt, which can be attributed to the difference of the bottom electrodes in work function and conductivity. The work function of Pt and TiN bottom electrodes is 5.65 and 4.5 eV, respectively; hence, the interface barrier between TiN and underlying HfO2 is relatively lower in the Pt/HfO2/TiO2/HfO2/TiN device. During a reset process, the needed energy for electron to cross the barrier is also smaller. So Pt/HfO2/TiO2/HfO2/TiN manifests lower resistance value in OFF state than Pt/HfO2/TiO2/HfO2/Pt correspondingly. During a set process, RRAM device changes from HRS into LRS owing to the formation of the conducting channels. The Pt bottom electrode has higher conductivity than TiN bottom one; accordingly, the resistance value in ON state for Pt/HfO2/TiO2/HfO2/TiN is higher than that for Pt/HfO2/TiO2/HfO2/Pt. As a result, Pt/HfO2/TiO2/HfO2/TiN devices show a smaller resistance ratio of OFF/ON states of 102. However, the resistance ratio of 102 has already met the requirement of RRAM applications.
In order to further understand the underlying resistive switching mechanism of HfO2/TiO2/HfO2 on Pt-coated and TiN-coated Si, we performed the XPS narrow scans and depth analyses on trilayer structures with symmetric and asymmetric bottom/top electrodes. XPS spectra were fitted with Gaussian-Lorentzian (G-L) functions after smart-type background subtraction.
The O 1s spectra from HfO2 and TiO2 layers can also be deconvoluted into two peaks, as seen in Fig. 6c, d. The relatively lower binding energies of the O 1s peak at ~530.9 and 531.2 eV are assigned to Ti-O and Hf-O bonding in TiO2 and HfO2 layers, respectively, belonging to lattice oxygen without oxygen vacancies. Whereas the slightly higher energy of 532.1 eV in the O 1s spectra of Fig. 6c, d are ascribed to the oxygen atoms near oxygen vacancies in HfO2 and TiO2 layers based on the literature reports [22, 23, 24, 25]. The relative oxygen vacancy concentration in the oxide layer can be roughly evaluated by calculating the area proportion of each peak [22, 23]. The calculated percentage concentration of oxygen vacancy in HfO2 and TiO2 layers is about 2.3 and 19.4%, respectively, in accord with the results of Hf n+ and Ti3+.
Figure 7c, d presents the distribution curves of oxygen vacancy concentration of HfO2/TiO2/HfO2 on Pt- and TiN-coated Si based on the above depth XPS profiles, respectively. The oxygen vacancy concentration was evaluated using the method mentioned above. It can be found that the highest oxygen vacancy concentration (~34%) appears in TiO2 intermediate layer. Another noteworthy feature is that the underlying HfO2 layer near Pt or TiN bottom electrodes has higher oxygen vacancy concentration than upper HfO2 layer (~2%). Besides, the oxygen vacancy concentration (~9%) of underlying HfO2 layer on TiN-coated Si is obviously higher than that (~6%) on Pt-coated Si.
Conversely, with a negative forming voltage on the top electrode, oxygen gas forms under the underlying HfO2 layer and above Pt the bottom electrode (Fig. 8e). The O2 releasing becomes more difficult, which will impede the formation of conductive filaments. Moreover, because the oxygen vacancy concentration of the upper HfO2 layer (~2.3%) is lower than that of the underlying HfO2 (~6%) (Fig. 7c), the higher negative forming voltage of −11 V is needed to form filaments. When more O2 accumulates to a certain pressure under underlying HfO2, it possibly erupts from the mechanically weakest part of the thin films, leading to the hole in oxide films or separation between oxide films and the bottom electrode induced by bubble cracking. Although the device can be reset to HRS at applied +4 V bias (Fig. 8g), the memory cell is degraded after several switching cycles. In our samples, a part of top electrode was blown off after the forming process. Similar electroforming polarity preference scenario has also been observed in Pt/TiO2−x /Pt bipolar RRAM cells with various physical deformation evidences of the junctions [31, 32].
Pt/HfO2/TiO2/HfO2/TiN devices exhibit quite different electroforming polarity from Pt/HfO2/TiO2/HfO2/Pt. The LRS caused by electroforming can be observed at both negative and positive bias voltage; however, the effective reset from LRS to HRS can be achieved only at positive bias voltage in Pt/HfO2/TiO2/HfO2/TiN device. Similar electroforming preference with asymmetric Pt and TiN electrodes has been observed in some literatures with various storage oxide layers such as HfO2 [33, 34], TiO2 , ZrO x /HfO y bilayer , and Al2O3/HfO2/Al2O3 trilayer . However, the related explanations are divergent or lacking.
For chemically inert Pt top electrode and relatively active TiN bottom electrode, when applying negative forming voltage, O2− ions drift towards the TiN anode and are absorbed to form TiO x N1−x layer (Fig. 9b), i.e., TiN + xO2− →TiO x N 1−x + x2e, avoiding the O2 release and possible damage to cells. The TiN electrode might act as a reservoir for oxygen ions drifting under an applied voltage [1, 36]. Simultaneously, due to the higher oxygen vacancy in the underlying HfO2 layer (~9%) and intermediate TiO2 layer (~10–34%) than the upper HfO2 layer (~1.9%), the migration of the rich oxygen vacancies from the underlying HfO2 layer through TiO2 layer towards the the upper HfO2 layer, directly producing conductive oxygen vacancy filaments with a smaller forming voltage of −3.7 V (Fig. 9b). When performing reverse reset process at +1.5 V, the bottom electrode of TiN layer acting as oxygen reservoir supplies oxygen ions to react with oxygen vacancy, i.e., O2− + V O 2+ →O O (lattice oxygen), which is beneficial to the reset operation from LRS to HRS with rupture of conductive filament (Fig. 9d). Above all, the oxygen reservoir effect of TiN bottom electrode contributes the forming voltage reduction as well as better set/reset voltage variation [1, 29, 36].
On the other hand, when imposing the positive forming voltage on top electrode Pt, the oxygen vacancies migrate from the upper HfO2 into the underlying HfO2 on TiN, accumulate in its vicinity, and develop into filaments. Owing to the lower oxygen concentration in the upper HfO2 (~1.9%) than the underlying HfO2 layer (~9%) and intermediate TiO2 layer (~10–34%), this leads to further increase of the oxygen vacancy concentration in underlying HfO2 and intermediate TiO2 layer. In the same time, due to slight oxygen existence in PEALD-derived TiN films (about 10%), some oxygen ions in TiN drift into underlying HfO2 layer to form insulate lattice oxygen with nearby oxygen vacancy, preventing from the growth of conductive filaments (Fig. 9e). Finally, when the bias voltage attains +14 V, the coarsening conductive filaments in oxides is formed. However, when a negative voltage is applied on the device, the conductive filament in trilayer structure is too large to be ruptured (Fig. 9g). Therefore, the device cannot switch to HRS by applying a negative voltage, indicating that an irreversible hard breakdown occurs in Pt/HfO2/TiO2/HfO2/TiN device.
Finally, on account of the fact that the inserted TiO2 layer stores more oxygen vacancies than HfO2 layer, the distribution of oxygen vacancies in trilayer structure is not uniform, especially in the two interfacial layers between HfO2/IL/TiO2/IL/HfO2, which might affect the growth position, direction, and overlapping of conductive filaments. The linkage or rupture of the conductive filaments corresponds to the set process from HRS to LRS or the reset process from LRS to HRS, respectively, which more easily happens in two interfacial layers. Further, the shape and position of the conductive filaments in HfO2 and TiO2 layers change less in the set and reset processes.
In summary, RRAM devices based on trilayer-structure of Pt/HfO2/TiO2/HfO2/Pt and Pt/HfO2/TiO2/HfO2/TiN have been prepared by ALD. Both memory cells show typical bipolar resistive switching characteristics, and Ohmic and SCLC dominant conduction mechanisms in LRS and HRS, respectively. It is found that the bottom electrodes of Pt and TiN have great influence on the electroforming polarity preference, the ratio of high and low resistances and dispersion of the operating voltage of trilayer-structure memory cells. Compared to with symmetric Pt top/bottom electrodes, the RRAM cells with asymmetric Pt top/TiN bottom electrodes show smaller negative forming voltage of −3.7 V, relatively narrow distribution of the set/reset operation voltages and lower ratio of high and low resistances of 102. The electrode-dependent electroforming polarity can be explained by considering electrodes’ chemical activity with oxygen, the related reactions at anode, and the nonuniform distribution of oxygen vacancy concentration in trilayer-structure of HfO2/TiO2/HfO2 on Pt- and TiN-coated Si. Furthermore, the TiN electrode as oxygen reservoir plays an important role in forming voltage reduction and better dispersion of RS parameters for Pt/HfO2/TiO2/HfO2/TiN devices. Considering the modulation effect of electrode and trilayer-structure on resistive switching performance, this work provides a new device design route for future RRAM applications.
This project is supported by the Natural Science Foundation of China and Jiangsu Province (51571111 and BK2016230), a grant from the State Key Program for Basic Research of China (2015CB921203). Ai-Dong Li also thanks the support of Priority Academic Program Development in the Jiangsu Province.
WZ, JZK, and ZYC carried out the sample fabrication and device measurements and drafted the manuscript. WZ, JZK, and ZYC analyzed the data and explained the results. WZ, JZK, and ZYC contributed equally to this work. LZ and XL helped to finish the X-ray photoelectron spectroscopy measurements. ADL, DW, and YQC participated in the discussion of results. ADL modified the manuscript and supervised all the projects. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 12.Gao B, Kang JF, Chen Y S, Zhang FF, Chen B, Huang P, Liu LF, Liu XY, Wang YY, Tran XA, Wang ZR, Yu HY, and Chin A. Oxide-based RRAM: unified microscopic principle for both unipolar and bipolar switching. IEDM Tech Dig. 2011;11:417–20.Google Scholar
- 19.Sherman A (2008) Atomic layer deposition for nanotechnology: an enabling process for nanotechnology fabrication. Ivoryton Press, New YorkGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.