The Switching Characteristics in Bilayer ZnO/HfO2 Resistive Random-Access Memory, Depending on the Top Electrode
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe article presents a well-structured investigation into the role of top electrodes (TE) in ZnO/HfO2 bilayer RRAM devices. The authors effectively combine experiment and theory, referencing established mechanisms of resistive switching and oxygen vacancy formation at oxide interfaces. The use of HfO2 and ZnO as switching and buffer layers, respectively, is well-supported by existing literature. The manuscript provides valuable insights into the relationship between TE properties and RRAM performance. Despite I am positively impressed by the experiment result, several critical points should be addressed to enhance the manuscript.
Below are the specific comments for reference:
1. Reviewer recommends the authors to include “Appl. Phys. Rev. 2022, 9, 021417; Chem. Rev. 2020, 120, 3941; Matter 2021, 4, 1702” as emerging memristor materials in the introduction.
2. Since the study investigates the variation of the top electrode material (TiN/Ti vs Pd), were the thicknesses of these top electrodes identical? If not, how could this difference influence the results?
3. The authors attribute the higher HRS resistance of the Pd TE device to its lower oxygen affinity compared to TiN/Ti. How could this difference in oxygen affinity be further investigated to confirm its role in influencing HRS resistance?
4. In the TiN/Ti TE device, why chooses ZnO as a buffer layer, providing oxygen ions or states of oxygen vacancy to other layers?
5. How do the authors explain the significant difference in cycling endurance observed between the TiN/Ti TE device and the Pd TE device? Can this difference be attributed solely to the material properties of the top electrodes (TE) or might other factors also play a role?
6. The passage suggests that the TiN/Ti TE device exhibits better stability than the Pd TE device under temperature and voltage stress. How do the authors relate this difference in stability to the size and number of conductive filaments formed in each device?
Author Response
1. Reviewer recommends the authors to include “Appl. Phys. Rev. 2022, 9, 021417; Chem. Rev. 2020, 120, 3941; Matter 2021, 4, 1702” as emerging memristor materials in the introduction.
- As you mentioned, we have incorporated the references as emerging memristor materials in the introduction. Various emerging memristor materials have recently been researched including metal-oxides, organic molecules, and 1D, 2D, and 0D materials [1-3]. The above modifications are highlighted in the “Introduction” section on page 1 of the manuscript.
2. Since the study investigates the variation of the top electrode material (TiN/Ti vs Pd), were the thicknesses of these top electrodes identical? If not, how could this difference influence the results?
- As you pointed out, the information on thickness is added to the “Materials and Methods” section on pages 2-3. A 10 nm of Ti and 50 nm of TiN were deposited, while Pd was 100 nm thick. It has been reported that a thicker top electrode (TE) provides greater leakage to the device [1-2]. In other words, the thicker TE has a lower value of HRS. However, in our research, the Pd TE device, which has a thicker TE than the TiN/Ti TE device, shows a lower leakage current and a large value of HRS. Therefore, we have concluded that the difference in TE thickness is negligible on the device performance.
[1] Rahaman, Sk Ziaur, et al. "The role of Ti buffer layer thickness on the resistive switching properties of hafnium oxide-based resistive switching memories." Langmuir 33.19 (2017): 4654-4665.
[2] He, Huikai, et al. "Ti/HfO2-Based RRAM with Superior Thermal Stability Based on Self-Limited TiOx." Electronics 12.11 (2023): 2426.
3. The authors attribute the higher HRS resistance of the Pd TE device to its lower oxygen affinity compared to TiN/Ti. How could this difference in oxygen affinity be further investigated to confirm its role in influencing HRS resistance?
- Thank you for pointing out the issues. Firstly, we assumed that oxygen ions would naturally migrate toward the TE after the fabrication. The higher the reactivity between the oxygen ions and the TE, the more oxygen ions will move, leading to an increase in oxygen vacancies within the oxide film. Therefore, in the case of Pd TE, due to its lower affinity for oxygen, there will be fewer oxygen vacancies in the oxide, resulting in a higher high resistance state (HRS). Similarly, we compared the HRS values for Pt TE and Ti TE in ZrO2 oxide in reference 23 and found that Pt TE consistently showed higher HRS values, which can be explained by the different extents to which oxygen vacancies are formed by the TE. Our experimental results also suggest that the lower affinity of Pd TE might result in fewer oxygen vacancies being formed, which would be the reason for its higher HRS compared to TiN/Ti TE. Moreover, reference 29 suggests that in-situ generated oxygen ions moving towards the Ti TE result in a lower HRS compared to devices with Pt TE, similar to those with Pd TE. Therefore, we have concluded that the lower HRS value observed in TiN/Ti TE devices, compared to those with Pd TE, is due to Ti's high oxygen affinity. As you mentioned, we have modified page 4 line 121 of the manuscript and highlighted it in red.
4. In the TiN/Ti TE device, why chooses ZnO as a buffer layer, providing oxygen ions or states of oxygen vacancy to other layers?
- ZnO is selected as the buffer layer because of its low binding energy, lower work function, and higher Gibbs free energy. Given its properties, ZnO effectively facilitates the provision of oxygen ions or states of oxygen vacancy to other layers. Additionally, oxygen ions inside ZnO also seem to be involved in the formation and disappearance of filaments. Considering your valuable feedback, we have updated the manuscript to better clarify the role of ZnO as a buffer layer in the “introduction” section on the 1 page of the manuscript.
5. How do the authors explain the significant difference in cycling endurance observed between the TiN/Ti TE device and the Pd TE device? Can this difference be attributed solely to the material properties of the top electrodes (TE) or might other factors also play a role
- Thank you for your feedback. The observed differences can be attributed solely to the material properties of the TE. Except for the material of the TE, all other processes are the same. Additionally, as noted in reference 13, unipolar operation typically exhibits less endurance than bipolar operation. This is because the Joule heating effect of the unipolar cycle influences the filaments. Therefore, we have concluded that the unexpected unipolar switching observed in the Pd device significantly reduces its cycling stability. We have accordingly updated the theoretical mechanism on page 8 of the manuscript.
6. The passage suggests that the TiN/Ti TE device exhibits better stability than the Pd TE device under temperature and voltage stress. How do the authors relate this difference in stability to the size and number of conductive filaments formed in each device?
- We apologize for the ambiguous statement in the manuscript. Firstly, the small number of filaments makes it probabilistically more susceptible to transitioning to the Low Resistance State (LRS) due to temperature-induced diffusion. Therefore, the quantity of filaments is a critical factor in stability. Also, the diffusion by joule heating is considered to stability. The Joule heating generated in each filament increases in proportion to the amount of current, as dictated by Equation (1).
J=I2Rt (1)
(where J is Joule heating, I is current, R is resistance, and t is time.)
The less current flows through each filament, allowing it to maintain a thermally stable state. This thermally stable state prevents the diffusion of oxygen ions, thus maintaining stable HRS under negative bias. Therefore, in the TiN/Ti TE device, smaller and many numbers of the size of the filaments are formed. Additionally, the larger number of filaments probabilistically contributes to robustness against the transition from high resistance state HRS to low resistance state LRS due to the diffusion of oxygen ions. In contrast, the Pd TE device has fewer, larger filaments formed. Therefore, we have proposed a filament model as shown in Figure 8. Considering your valuable feedback, we have modified the manuscript in the “Results and Discussion” section on page 7 and page 8 of the manuscript.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe article "The Switching Characteristics in Bilayer ZnO/HfO2 Resistive Random-Access Memory Depending on Top Electrode" provides a comprehensive analysis of bipolar switching behaviors in ZnO/HfO2 bilayer RRAM with different top electrodes. This study examined the impact of different electrodes on electrical properties. Although this topic is not new and is well-explored, the approach using Gibbs free energy to explain electrical behaviors across different layers is noteworthy. The study details how variations in Gibbs free energy across these layers help in understanding the movement of oxygen ions, which in turn explains the operating voltages and high resistance state (HRS) magnitudes of various devices. The differences in Gibbs free energy across layers elucidate the ease of oxygen ion migration, thereby explaining the operating voltages and HRS magnitudes of different devices. It also explains the impact of the Ti electrode forming TiOx at the interface on the oxygen vacancies in in the neighboring ZnO layer, thereby clarifying several observed phenomena.
However, the article has several areas that require attention:
1. The device exhibits a positive bias voltage for SET and a negative bias voltage for RESET. Figure 7(b) shows increasing instability with greater negative bias, leading to a change from HRS to LRS. Contrarily, a negative voltage drives oxygen ions towards the HfO2 layer (as shown in Fig.4(b)), reducing oxygen vacancies within HfO2 and disrupting the filament paths (as illustrated in Fig.8(f)). This indicates a contradiction between the phenomena observed in Fig.7(b) and the theoretical explanation provided in Fig.4(b).
2. The numbering of figures in Fig. 7 does not match the text, leading to confusion that needs correction.
3. Figure 9 appears to be wrong or unclear. The illustration of single layer, active electrode, and inactive electrode appears to be inconsistent with the main text and previous phenomena. It will cause readers to misunderstand the phenomenon and explanation.
4. The author proposes that differences in work function contribute to stability variations but fails to clarify this point. They reference studies like ref.27, which uses work function to describe different transition states but overlooks the Gibbs free energy differences between AlOx and PtOx. It could use Gibbs free energy differences to explain the ease of oxygen ion movement, which could explain how the Ti electrode improves stability. If the work function is considered the primary factor, a more thorough and compelling explanation should be provided.
This article could contribute to the field after major revision. In its present form, the manuscript is not ready for publication and requires significant improvements and a more comprehensive explanation from the authors before it can be reconsidered.
Author Response
1. The device exhibits a positive bias voltage for SET and a negative bias voltage for RESET. Figure 7(b) shows increasing instability with greater negative bias, leading to a change from HRS to LRS. Contrarily, a negaive voltage drives oxygen ions towards the HfO2 layer (as shown in Fig.4(b)), reducing oxygen vacancies within HfO2 and disrupting the filament paths (as illustrated in Fig.8(f)). This indicates a contradiction between the phenomena observed in Fig.7(b) and the theoretical explanation provided in Fig.4(b).
- We apologize for the ambiguous statement in the manuscript. In the ideal theoretical models in Fig. 4(b) and Fig. 8 (f), the models do not account for diffusion by Joule heating, leading to the expectation that filaments should break only depending on a negative bias. However, the actual conditions of Fig.7 (b) result in high internal temperatures that cause filaments to dissolve in a somewhat random manner. This high thermal activity disrupts the filaments unpredictably, contributing to the instability under negative biases as seen in Figure 7 (b). It appears there was a lack of clarity in our theoretical explanation, which may have caused confusion. We have revised Figure 4 and Figure 8 in our manuscript to provide a clearer understanding of the underlying mechanisms.
2. The numbering of figures in Fig. 7 does not match the text, leading to confusion that needs correction.
- As you pointed out, the numbering of figures in Fig. 7 did not match the text, leading to confusion. Accordingly, we have corrected the figure numbering.
3. Figure 9 appears to be wrong or unclear. The illustration of single layer, active electrode, and inactive electrode appears to be inconsistent with the main text and previous phenomena. It will cause readers to misunderstand the phenomenon and explanation.
-Thank you for pointing out the issues with Figure 9. To reduce confusion, we have removed the depiction of the single layer. Additionally, we have corrected the inconsistencies regarding each electrode to clarify the phenomena more accurately in Figure 8. We appreciate your feedback and hope these changes improve the understanding of the figure's context of the text.
4. The author proposes that differences in work function contribute to stability variations but fails to clarify this point. They reference studies like ref.27, which uses work functions to describe different transition states but overlooks the Gibbs free energy differences between AlOx and PtOx. It could use Gibbs free energy differences to explain the ease of oxygen ion movement, which could explain how the Ti electrode improves stability. If the work function is considered the primary factor, a more thorough and compelling explanation should be provided.
-We apologize for the ambiguous statement in the manuscript regarding the lack of explanation of the theoretical mechanism related to the work function. Considering your valuable feedback, we have concluded the work function is not considered a major factor, and oxygen movement seems to be explainable with Gibbs free energy. We have removed the explanation of page 8 on the manuscript and have modified sentences related to the theory of work function in the 'Abstract,' 'Introduction,' and 'Conclusion' sections.
Reviewer 3 Report
Comments and Suggestions for Authors1. This introduction does not provide a comprehensive review of the extensive knowledge of the top electrode (TE) used in RRAM. It does not present either to the readers the challenges remaining to be tackled and the work already done that has been done using TiN/Ti TE or Pd TE in RRAM. The motivation of the work is therefore unclear and the novelty of the work not demonstrated.
2. The detail preparation process of ALD for HfO2 and ZnO layers should be descripted in the Materials and Methods.
3. In Materials and Methods, some details could be added such as RF-sputtering process, XRD measurement and electrical characteristics.
4. In Fig. 3, “HRS” and “LRS” should be marked.
5. In page 3 line 81-82, I cannot well understand the meaning of “the high oxygen reactivity of the TiN/Ti TE is attributed to an increase in oxygen vacancies, resulting in a relatively low resistance value in HRS “. The reason why low resistance value in HRS for TiN/Ti TE should be better explored.
Comments on the Quality of English LanguageThe English in the passage is generally clear and technically sound, but some minor revisions could enhance clarity and precision.
Author Response
1. This introduction does not provide a comprehensive review of the extensive knowledge of the top electrode (TE) used in RRAM. It does not present either to the readers the challenges remaining to be tackled and the work already done that has been done using TiN/Ti TE or Pd TE in RRAM. The motivation of the work is therefore unclear and the novelty of the work not demonstrated.
- As you pointed out, we added a comprehensive review of TE to provide extensive knowledge. However, the stable bipolar operation is influenced not only by the oxide resistance layer but also by the top electrode (TE). In particular, bipolar operation tends to have longer endurance compared to unipolar operation due to its different reset mechanism [13], highlighting the critical role of TE. In the simulated results, both the unipolar and bipolar switching mechanisms in RRAM, depending on the TE’s varying oxygen affinity, have been reported [14]. In unipolar switching, Joule heating is a primary factor in the reset operation. The Joule heating generated by the current thermally activates the diffusion of oxygen ions, which typically migrate from the interface or the area surrounding the filaments [15]. The previous research reported that the local temperature around the filaments would rise to several hundred Kelvin due to the large current flow, as determined by electrothermal calculations [16]. The unipolar operation occurs when TE is inert materials such as Pd or Pt. In contrast, bipolar switching occurs when the TE is oxidizable. The TE is oxidizable materials such as Ti or TiN, and the bipolar operation occurs [17]. M. Uenuma [18] has analyzed Joule heating in the NiO RRAM, reporting that bipolar operation generates less Joule heating, whereas unipolar operation produces significantly more heat. Considering these factors, the thermal situation and stability within the ZnO/HfO2 bilayer oxide are interconnected and different depending on the TE. Nevertheless, research on the stability of bipolar switching in ZnO/HfO2 bilayer RRAM remains insufficient. The above modifications are highlighted in red in the “Introduction” section of the manuscript.
Additionally, in ZnO/HfO2 bilayer RRAM, research about stability depending on the effect of TE is insufficient. In response to your thoughtful feedback, we have further clarified the novelty of our work in the "Introduction," "Abstract," and "Conclusion" of our manuscript, and highlighted in red.
2. The detail preparation process of ALD for HfO2 and ZnO layers should be descripted in the Materials and Methods.
- As you mentioned, we added the ALD process in detail. Subsequently, layers of 3 nm HfO2 and 3 nm ZnO are deposited by atomic layer deposition (ALD). Tetrakis (dimethylamino) hafnium (TDMAHf) and H2O is used as the precursor for HfO2 at 200 °C and diethylzinc (DEZ) and H2O is used for the ZnO layer at 80 °C. The above modifications are highlighted in the "Materials and Methods" section on page 2 in red.
3. In Materials and Methods, some details could be added such as RF-sputtering process, XRD measurement and electrical characteristics.
- Thank you for your valuable suggestions. In response to your comments, we have updated the "Materials and Methods" section as follows:
RF-sputtering process: Then, a Pd/Ti bottom electrode (BE) is deposited using an RF-sputter (KVS-2000 series, Korea Vacuum Tech, Gimpo-si, Korea). This sputtering system operates at a base pressure of 2 ×10-6 Torr with an argon gas flow of 10 sccm and a working pressure of 2 mTorr. The targets used are Ti (> 99.95 %) and Pd (> 99.95 %), with a sputtering power of 100 W.
XRD measurement: The crystal phase and grain size are analyzed using X-ray diffraction (XRD, SmartLab 9 kW, Rigaku Corporation, Tokyo, Japan), which is equipped with a Cu-Kα radiation instrument (λ = 1.54 Å) in the range of 2θ = 10–90°.
Electrical Characteristics: To investigate the switching characteristics, the DC I-V electrical characteristics and voltage stress are measured using a semiconductor analyzer (HP4155B, HP, Englewood, CO, USA) equipped with a static source measure unit (SMU). Resistance measurements at various temperatures are performed on a hot stage (Linkam LTS 420, Linkam Scientific Instruments Ltd., Tadworth, Surrey, UK). During the measurements, bias is applied to the TE while the BE is grounded.
The above modifications are highlighted in the "Materials and Methods" section on page 2 in red, for each process.
4. In Fig. 3, “HRS” and “LRS” should be marked.
- As you mentioned, we have marked the “HRS” and “LRS” in Fig. 3.
5. In page 3 line 81-82, I cannot well understand the meaning of “the high oxygen reactivity of the TiN/Ti TE is attributed to an increase in oxygen vacancies, resulting in a relatively low resistance value in HRS “. The reason why low resistance value in HRS for TiN/Ti TE should be better explored.
- Thank you for pointing out the issues. Firstly, we assumed that oxygen ions would naturally migrate toward the TE after the fabrication. The higher the reactivity between the oxygen ions and the TE, the more oxygen ions will move, leading to an increase in oxygen vacancies within the oxide film. Therefore, in the case of Pd TE, due to its lower affinity for oxygen, there will be fewer oxygen vacancies in the oxide, resulting in a higher high resistance state (HRS). Similarly, we compared the HRS values for Pt TE and Ti TE in ZrO2 oxide in reference 23 and found that Pt TE consistently showed higher HRS values, which can be explained by the different extents to which oxygen vacancies are formed by the TE. Our experimental results also suggest that the lower affinity of Pd TE might result in fewer oxygen vacancies being formed, which would be the reason for its higher HRS compared to TiN/Ti TE. Moreover, reference 29 suggests that in-situ generated oxygen ions moving towards the Ti TE result in a lower HRS compared to devices with Pt TE, similar to those with Pd TE. Therefore, we have concluded that the lower HRS value observed in TiN/Ti TE devices, compared to those with Pd TE, is due to Ti's high oxygen affinity. As you mentioned, we have modified page 4 of the manuscript and highlighted it in red.
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have substantially revised the manuscript and can reasonably explain their experimental results. However, the numbering in Fig. 7, which they claim to have corrected, has not been corrected yet. Please check this carefully. After correction, I recommend that this manuscript be accepted for publication.
Author Response
Dear Reviewer,
We sincerely apologize for the oversight regarding the numbering in Fig. 7. It appears that there was an error during the transcription of the revised manuscript. We have now identified and carefully corrected this issue. Thank you for your valuable feedback and for bringing this to our attention. We have updated the manuscript accordingly and hope it now meets the requirements for publication.
Reviewer 3 Report
Comments and Suggestions for AuthorsThe revisions have effectively addressed the initial concerns and suggestions, enhancing the overall clarity and impact of the manuscript. Therefore, it is recommended for acceptance in its present form.
Author Response
Dear Reviewer,
Thank you for your positive feedback. We are pleased to hear that the revisions have effectively addressed the initial concerns and suggestions, enhancing the overall clarity and impact of the manuscript. We appreciate your recommendation for acceptance in its present form.