- Email: [email protected]
Revealing multimode resistive switching in Cu-O nanostructures using conductive atomic force microscopy
Accepted Manuscript Full Length Article Revealing multimode resistive switching in Cu-O nanostructures using conductive atomic force microscopy Mohit Kumar, Biswarup Satpati, Tapobrata Som PII: DOI: Reference:
S0169-4332(18)31449-1 https://doi.org/10.1016/j.apsusc.2018.05.137 APSUSC 39413
To appear in:
Applied Surface Science
Received Date: Revised Date: Accepted Date:
15 November 2017 10 April 2018 18 May 2018
Please cite this article as: M. Kumar, B. Satpati, T. Som, Revealing multimode resistive switching in Cu-O nanostructures using conductive atomic force microscopy, Applied Surface Science (2018), doi: https://doi.org/ 10.1016/j.apsusc.2018.05.137
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Revealing multimode resistive switching in Cu-O nanostructures using conductive atomic force microscopy Mohit Kumar,1,2,† Biswarup Satpati,2,3 and Tapobrata Som1,2,* 1
SUNAG Laboratory, Institute of Physics, Sachivalaya Marg, Bhubaneswar 751 005, India
2
Homi Bhabha National Institute, Training School Complex, Anushakti Nagar,
Mumbai 400 085, India 3
Surface Physics and Materials Science Division, Saha Institute of Nuclear Physics, 1/AF
Bidhannagar, Kolkata 700 064, India Abstract We demonstrate the nanoscale multimode resistive switching in copper oxide nanostructures using conductive atomic force microscopy (cAFM). The cross-sectional transmission electron microscopy (XTEM) and scanning tunnelling electron microscopy – high angle angular dark field (STEM-HADDF) imaging confirms the formation of Cu-O nanostructures. In addition, xray photoelectron spectroscopy (XPS) is used to study the chemical composition of Cu-O nanostructures. Current-voltage characteristics measured by conductive atomic force microscopy (cAFM) reveals that the filament forms in multistep processes, instead the rapid one, indicating the multimode resistive switching. The presence of multimode RS is corroborated to the defectinduced conduction mechanism across the Cu-O nanostructures. The present study should open up a new avenue to understand the conduction mechanism and to design an advanced nanoscale device. *Author to whom correspondence should be addressed; E-mail: [email protected] †Present address: Department of Electrical Engineering, Incheon National University, 119 Academy Rd. Yeonsu, Incheon, 406772, Republic of Korea 1
1. Introduction Future electronics demand a better control on performance, variability, and reliability of electrically-triggered resistive switching (RS) for its use in non-volatile memory devices [1]. The memory effect is explained in terms of nanoscale conductive filament formation and its rupture embedded in an insulating matrix [2]. Generally, during the RS, atomic rearrangement takes place which in turn makes it more complicated in comparison to traditional memories, especially, at the nanoscale [2,3]. In fact, as memory devices shrink towards the nanoscale, quantum mechanics comes into the picture and thus, in order to realize the full potential of such devices, understanding the filament formation and charge transport across it become exigent. However, due to the limitations involved with the present bulk current measurement techniques, understanding the origin of RS and its material-dependent deviating behaviour are still considered to be fundamental but unsolved riddles. In this regard, it is interesting to note that the versatile potential of scanning probe microscopy (SPM) has made it suitable to study several nanoscale physical properties of materials [4]. Among different modes of SPM, conductive atomic force microscopy (cAFM) is used extensively because it can build up simultaneous information on local electrical conduction and surface topography [5]. In fact, a one-to-one correspondence between the surface morphology and current mapping, with a very high lateral resolution (~5 nm in topography and ~1 pA in current image), can be achieved through cAFM measurements. In addition, in cAFM, a conductive tip is used as a top electrode which gives an easy control to perform I–V measurements at a desired location [4,5]. Among various RS materials, nontoxic, naturally abundant binary phases of copper oxides, such as cuprous oxide (Cu2O), cupric oxide (CuO), and CuOx have attracted considerable attention not only for easy processing of efficient and compact resistive random-
2
access memory (ReRAM) by scaling down the device structure but also for their full compatibility with standard copper [5-8]. Most of the prior results indicate that a mixed phase of Cu2O and CuO (e.g. Cu-O) is more useful for the stability of a device [8]. In fact, RS behaviour of bulk and thin film Cu-O are studied by various groups which demonstrate an excellent retention time [5,7,8,9]. However, with a decreasing trend in the size of the memory devices (down to nanoscale), filament formation and its rupture will be confined within nanometer, which is yet to be explored. Thus, in order to design an efficient and a controlled device, it is necessary to have a better understanding on the conduction mechanism responsible for the RS behaviour across a nanoscale Cu-O film. On the other hand, to the best of our knowledge, the same are not fully explored yet. In this paper we demonstrate the presence of multimode resistive switching in Cu-O nanostructured thin film and the associated current conduction mechanism. The particle size distribution and elemental analysis are carried out using atomic force microscopy (AFM), crosssectional transmission electron microscopy (XTEM) and scanning tunnelling electron microscopy – high angle angular dark field (STEM-HADDF) imaging. In addition, x-ray photoelectron spectroscopy (XPS) is used to study the chemical composition of Cu-O nanostructures. Current-voltage characteristics measured using conductive atomic force microscopy (cAFM) reveal that filaments form in multistep processes, instead of the rapid one, indicating the presence of multimode resistive switching. The presence of multimode RS corroborates with defect-induced conduction mechanism across the Cu-O nanostructures. The present study should open up a new avenue to design a Cu-O-based advanced nanoscale memory device. 2. Experimental
3
Nanostructures of copper oxide, having a nominal thickness of 10 nm, was grown at room temperature (RT) on ultrasonically cleaned low resistive (ρ=1-510-3 Ω-cm) p-Si(100) wafers (area 11 cm2) by RF magnetron sputtering technique. The base pressure of the chamber (Excel Instruments, India) was 5×10-7 mbar and during the deposition of copper oxide film and the argon gas flow rate was maintained at 30 sccm, resulting in the working pressure of 5×10-3 mbar. The distance from the sample to the target was fixed at 80 mm and an RF power of 100 W was applied on the target. Microstructural analysis of copper oxide film was carried out by crosssectional transmission electron microscopy (XTEM) using a high-resolution transmission electron microscopy (HRTEM) ( FEI, Tecnai G2 F30-ST) operating at 300 kV equipped with a CCD camera (Gatan Orius). High-angle annular dark field (HAADF) scanning transmission electron microscopy (STEM) was employed using the same microscope, which was equipped with a scanning unit and a HAADF detector (Fischione, Model 3000). The compositional analysis was performed by energy dispersive X-ray spectroscopy (EDS) attachment (EDAX Inc.) on the Tecnai G2 F30. The compositional analysis of the Cu-O thin film was carried out using xray photoelectron spectroscopy (XPS) (VG Instruments). Further, the localized electrical transport phenomenon for forward and reverse voltage sweep of 10 V was investigated ex-situ by conductive atomic force microscopy (cAFM) (Asylum Research, MFP-3D). Conductive (Olympus, AC240TM) tip having ∼30 nm radius of curvature, ∼2 Nm−1 stiffness, and a resonance frequency of ∼70 kHz was used for electrical measurement. All recorded AFM images were analyzed using the WSxM software [10].
3. Results and discussion
4
Prior to exploring the RS behaviour in copper oxide, the sample is characterized by TEM and XPS. Figure 1(a) shows the cross-sectional TEM image of the copper oxide nanostructured film deposited on a Si substrate which clearly depicts the presence of uniformly distributed nanostructures, having an average size of 12±2 nm. Figure 1(b) shows a magnified view of the selected region marked on Fig. 1(a). A HRTEM image, collected from the marked region on Fig. 1(b) is presented in Fig. 1(c) which reveals the crystalline nature of Cu-O nanostructures. In fact, the d-spacing, calculated from this lattice image, matches well with the (111) plane of CuO phase. Likewise, HRTEM image collected from large randomly chosen places on the sample (images not shown) and d-spacings calculated from those matches closely with those of CuO and/or Cu2O phases. Thus, it becomes evident that both the phases (CuO and Cu2O) are present in the Cu-O nanostructured thin film. In addition, STEM-HAADF measurements confirm the presence of copper, oxygen, and silicon in the sample [Figs. 1(d)-(g)]. In addition, the morphology of Cu-O nanostructured thin film is studied by AFM measurement [Fig. 2(a)]. Here the inset depicts the grain size distribution estimated from this image where the average grain size turns out to be 12 nm. The compositional analyses are carried out by XPS. For instance, the high-resolution core level spectrum of Cu 2p is depicted in Fig. 2(b) [11] where the two peaks located at around 934.32 and 953.72 eV are attributed to the binding energy of Cu 2p3/2 and Cu 2p1/2, respectively, revealing an oxidation state of +2 [12]. Moreover, the two extra satellite peaks are also observed on a higher binding energy side, 942.0 and 962.2 eV for Cu 2p3/2 and Cu 2p1/2, respectively, implying the presence of an unfilled Cu 3d 9 shell and thus further confirms the existence of Cu2+ on the sample surface [13]. From the peak fitting of Cu 2p, we find two other small peaks, with binding energy of 932.6 and 952.6 eV, which are assigned to Cu 2p3/2 and Cu 2p1/2 in Cu2O, respectively [12,13]. The occurrence of a
5
satellite feature on the higher binding energy (943.2 eV) side of the Cu 2p main peak indicates the presence of CuO, because cupric oxide (Cu2+) has hole states in the Cu 3d band (Cu 3d9 configuration). This result indicates the deoxidization process and confirms the coexistence of CuO and Cu2O phases in the nanostructures (as discussed above). On the other hand, Fig. 2 (c) shows the XPS spectra corresponding to O 1s core level which can be deconvoluted into two Gaussian peaks, after Shirley background subtraction [14]. The broad and asymmetric nature of the peak can be ascribed to multiple co-ordinations of oxygen in the Cu-O nanostructures. The deconvoluted peaks at a binding energy of 529.0 and 531.8 eV are attributed to oxygen in Cu(OH)2 and Cu2O, respectively [15]. Figure 3(a) shows the local I-V characteristics of nanostructures, using cAFM measurements, obtained by sweeping a DC bias according to 0 V → 10 V → 0 V → -10 V → 0 V with a ramp rate of 0.2 V s-1 at a fixed tip location on a randomly chosen topographical places. It is noted that nanostructures exhibit bipolar switching behavior: it sets at positive bias and resets at a negative one corresponding to the current compliance of 20 nA. The I-V curve, presented in Fig. 3(a) is separately analyzed for different branches: at positive voltages, the current is very low and remains constant (path #1) until the voltage reaches around +7.5 V from where the current increases suddenly (path #2) and the corresponding voltage is identified as the ‘set’ voltage (Vset). It is interesting to note that during the set processes highly reproducible multiple current jumps are detected. This particular feature is not generic and is observed at wide range of topographical locations. Beyond 7.5 V, the current value reaches the compliance limit (20 nA) of the cAFM setup and hence, shows a straight line (path #3). On the other hand, during the decreasing cycle of voltage, current remains saturated till far below Vset (7.5 V). At much lower voltages (<3 V), current decreases almost linearly, indicating the Ohmic nature of the tip-
6
sample contact (path #4). A small deviation from the ohmic behaviour is due to the presence of bottom Si substrate. It is worth mentioning here that I-V characteristics are recorded from a large number of randomly chosen points on the sample to observe that behaviour of the curves remains nearly similar. A pronounced change in resistance is also observed for both increasing and decreasing voltages, corresponding to a tip bias (Vtip) of 5 V which is known as high resistance state (HRS) and low resistance state (LRS), respectively. Subsequently, an opposite ‘reset’ process is also observed when sweeping the voltage from 0 to -6 V (Vreset, path #5), as is evidenced by a two-step switching from LRS to HRS [Fig. 3(a)]. Figure 3(b) shows the forward bias set I–V curves in the forming voltage region in semilog scale. For a better clarity, a zoomed part of Fig. 3(b) is also depicted in Fig. 3(c). It is interesting to note that during the SET process, current does not increase suddenly but instead it goes through steps. This indicates that a particular process is not responsible for the filament formation and instead multiple-mode processes take place during the same. The exact multimode mechanism during the set process is still under investigation, yet the possible reason behind this may be a sudden increase in the current which can generate heat within the nanostructures and in turn induce defect injection, resulting an improvement in the filament size [Fig. 4(a)] [16,17,18,19]. In fact, the size of the filament can change due to defect injection/extraction during the set process [Fig. 4(b)]. The observed effect is quite significant since it suggests that the defects are collected by applied voltage. On the other hand, the reset process is attributed to the thermal dissolution of the metal filament due to Joule-heating-assisted oxidation followed by the diffusion of metal ions under the concentration gradient and the applied electric field (the ON and OFF states are schematically illustrated in the insets) [Fig. 4(c)]. The migration of oxygen vacancies can be described by two mechanisms: (i) drift and (ii) diffusion [19,20]. The driving
7
forces for drift and diffusion are the gradients of the electrostatic and electrochemical potentials, respectively. The diffusion coefficient and mobility are the most important parameters essential for describing the voltage-dependent migration behaviour of charged particles including oxygen ions and corresponding defects. Such pathways are around 5–30 nm in diameter, suggesting that devices may be scaled down to the nanometer dimensions to achieve very high levels of integration.
4. Conclusions In conclusion, we have observed multiple level SET process in the RS phenomenon in Cu-O nanostructures. AFM, XTEM, STEM-HADDF, and XPS studies are employed to evaluate sample morphology, extract particle size distribution, study the microstructure, and find out the atomic compositions. It is observed that crystalline Cu-O nanostructured thin film grown at RT shows the coexistence of both CuO and Cu2O phases. cAFM measurements reveal that filament formation takes place in multistep processes, instead of a rapid one, confirming the multimode resistive switching. The observed results are explained in terms of thermally-induced defect injection and migration and in turn improvement in the filament size. The model suggests that the optimization of the conductive filament active region is key for the future nanoscale resistive switching devices. The present results will be a benchmark to design efficient resistive Cu-O nanostructure-based memory devices.
Acknowledgements The authors thank to Professor Shikha Varma and Mr. Santosh Kumar Choudhury from Institute of Physics, Bhubaneswar for the XPS measurements.
8
References [1] K. M. Kim, D. S. Jeong, and C. S. Hwang, Nanotechnology 22 (2011) 254002. [2] R. Waser, R. Dittmann, G. Staikov, and K. Szot, Adv. Mater. 21 (2009) 2632. [3] Mario Lanza, Materials 7 (2014) 2155. [4] E. Meyer, H. J. Hug, R. Bennewitz Scanning Probe Microscopy. The Lab on a Tip SpringerVerlag, Heidelberg, Germany (2004). [5] M. Kumar and T. Som Nanotechnology 26 (2015) 345702. [6] K. Park and J.-S. Lee Nanotechnology 27 (2016) 125203. [7] A. Chen, S. Haddad, Y. C. Wu, Z. Lan, T. N. Fang, and S. Kaza, Appl. Phys. Lett. 91 (2007) 123517. [8] S. Y. Wang, C. W. Huang, D. Y. Lee, T. Y. Tseng, and T. C. Chang, J. Appl. Phys. 108 (2010) 114110. [9] J. W. Han and M. Meyyappan, AIP Adv. 1 (2011) 032162. [10] http://www.nanotec.es/products/wsxm/ [11] J. P. Espinos, J. Morales, A. Barranco, A. Caballero, J. P. Holgado, and A. R. GonzalezElipe, J. Phys. Chem. B 106 (2002) 6921. [12] C.-K. Wu, M. Yin, S. O’Brien, and J. T. Koberstein, Chem. Mater. 18 (2006) 6054. [13] J. Morales, J. P. Espinos, A. Caballero, A. R. Gonzalez-Elipe, J. A. Mejias, J. Phys Chem B. 109 (2005) 7758.
9
[14] Y. Wang, S. Lany, J. Ghanbaja, Y. Fagot-Revurat, Y. P. Chen, F. Soldera, D. Horwat, F. Mucklich, and J. F. Pierson, Phys. Rev. B 94 (2016) 245418. [15] J. Morales, L. Sanchez, F. Martın, J. R. Ramos-Barrado, M. Sanchez, Thin Solid Films 474 (2005) 133. [16] H. D. Lee, B. Magyari-Köpe, and Y. Nishi, Phys. Rev. B 81 (2010) 193202. [17] S. Nigo, M. Kubota, Y. Harada, T. Hirayama, S. Kato, H. Kitazawa, and G. Kido, J. Appl. Phys. 112 (2012) 033711. [18] U. Celano, G. Giammaria, L. Goux, A. Belmonte, M. Jurczaka and W. Vandervorsta, Nanoscale, 8 (2016) 13915. [19] K.-C. Kwon, M.-J. Song, K.-H. Kwon, H.-V. Jeoung, D-W. Kim, G-S. Lee, J.-P. Hong and J.-G. Park, J. Mater. Chem. C, 3 (2015) 9540. [20] Y. Hou, U. Celano, L. Goux, L. Liu, R. Degraeve, Y. Cheng, J. Kang, M. Jurczak, and W. Vandervorst, Appl. Phys. Letts. 109 (2016) 023508. [21] Yuchao Yang and Wei Lu, Nanoscale, 5 (2013) 10076.
10
List of figures FIG. 1. (Color online) (a) The XTEM image of Cu-O nanostructures. (b) Shows the magnified view of the marked region on (a). (c) Shows a HRTEM image obtained from the marked region on (b). STEM-HAADF elemental map of Cu, O and Si, and total are shown in (d)-(g), respectively.
FIG. 2. (Color online) (a) AFM image of the as-deposited Cu-O thin film on a Si substrate. The inset shows the particle size distribution (b) and (c) Depict Cu 2p and O 1s XPS spectra, respectively obtained from the Cu-O nanostructured film.
FIG. 3. (Color online) (a) Local I-V characteristics performed at various points, whereas the semi-log of the forward bias for the same is depicted in (b). For a clarity, zoomed part of the (b) shows in (c).
FIG. 4. (Color online) Schematic diagrams of filament formation and its growth are depicted in (a) and (b), respectively. (c) reset process, presenting the filament breakdown.
11
Figure 1 Kumar et al.
12
Figure 2 Kumar et al.
13
Figure 3 Kumar et al.
14
Figure 4 Kumar et al.
15
Highlights
1. Depicting nanoscale multimode resistive switching (RS) in nanostructured copper oxide thin film 2. Utilization of conductive atomic force microscopy for studying this nanoscale behavior 3. XTEM and STEM-HADDF imaging confirms the formation of Cu-O nanostructures 4. RS corroborates with defect-induced conduction mechanism
16
Graphical abstract
17
S0169-4332(18)31449-1 https://doi.org/10.1016/j.apsusc.2018.05.137 APSUSC 39413
To appear in:
Applied Surface Science
Received Date: Revised Date: Accepted Date:
15 November 2017 10 April 2018 18 May 2018
Please cite this article as: M. Kumar, B. Satpati, T. Som, Revealing multimode resistive switching in Cu-O nanostructures using conductive atomic force microscopy, Applied Surface Science (2018), doi: https://doi.org/ 10.1016/j.apsusc.2018.05.137
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Revealing multimode resistive switching in Cu-O nanostructures using conductive atomic force microscopy Mohit Kumar,1,2,† Biswarup Satpati,2,3 and Tapobrata Som1,2,* 1
SUNAG Laboratory, Institute of Physics, Sachivalaya Marg, Bhubaneswar 751 005, India
2
Homi Bhabha National Institute, Training School Complex, Anushakti Nagar,
Mumbai 400 085, India 3
Surface Physics and Materials Science Division, Saha Institute of Nuclear Physics, 1/AF
Bidhannagar, Kolkata 700 064, India Abstract We demonstrate the nanoscale multimode resistive switching in copper oxide nanostructures using conductive atomic force microscopy (cAFM). The cross-sectional transmission electron microscopy (XTEM) and scanning tunnelling electron microscopy – high angle angular dark field (STEM-HADDF) imaging confirms the formation of Cu-O nanostructures. In addition, xray photoelectron spectroscopy (XPS) is used to study the chemical composition of Cu-O nanostructures. Current-voltage characteristics measured by conductive atomic force microscopy (cAFM) reveals that the filament forms in multistep processes, instead the rapid one, indicating the multimode resistive switching. The presence of multimode RS is corroborated to the defectinduced conduction mechanism across the Cu-O nanostructures. The present study should open up a new avenue to understand the conduction mechanism and to design an advanced nanoscale device. *Author to whom correspondence should be addressed; E-mail: [email protected] †Present address: Department of Electrical Engineering, Incheon National University, 119 Academy Rd. Yeonsu, Incheon, 406772, Republic of Korea 1
1. Introduction Future electronics demand a better control on performance, variability, and reliability of electrically-triggered resistive switching (RS) for its use in non-volatile memory devices [1]. The memory effect is explained in terms of nanoscale conductive filament formation and its rupture embedded in an insulating matrix [2]. Generally, during the RS, atomic rearrangement takes place which in turn makes it more complicated in comparison to traditional memories, especially, at the nanoscale [2,3]. In fact, as memory devices shrink towards the nanoscale, quantum mechanics comes into the picture and thus, in order to realize the full potential of such devices, understanding the filament formation and charge transport across it become exigent. However, due to the limitations involved with the present bulk current measurement techniques, understanding the origin of RS and its material-dependent deviating behaviour are still considered to be fundamental but unsolved riddles. In this regard, it is interesting to note that the versatile potential of scanning probe microscopy (SPM) has made it suitable to study several nanoscale physical properties of materials [4]. Among different modes of SPM, conductive atomic force microscopy (cAFM) is used extensively because it can build up simultaneous information on local electrical conduction and surface topography [5]. In fact, a one-to-one correspondence between the surface morphology and current mapping, with a very high lateral resolution (~5 nm in topography and ~1 pA in current image), can be achieved through cAFM measurements. In addition, in cAFM, a conductive tip is used as a top electrode which gives an easy control to perform I–V measurements at a desired location [4,5]. Among various RS materials, nontoxic, naturally abundant binary phases of copper oxides, such as cuprous oxide (Cu2O), cupric oxide (CuO), and CuOx have attracted considerable attention not only for easy processing of efficient and compact resistive random-
2
access memory (ReRAM) by scaling down the device structure but also for their full compatibility with standard copper [5-8]. Most of the prior results indicate that a mixed phase of Cu2O and CuO (e.g. Cu-O) is more useful for the stability of a device [8]. In fact, RS behaviour of bulk and thin film Cu-O are studied by various groups which demonstrate an excellent retention time [5,7,8,9]. However, with a decreasing trend in the size of the memory devices (down to nanoscale), filament formation and its rupture will be confined within nanometer, which is yet to be explored. Thus, in order to design an efficient and a controlled device, it is necessary to have a better understanding on the conduction mechanism responsible for the RS behaviour across a nanoscale Cu-O film. On the other hand, to the best of our knowledge, the same are not fully explored yet. In this paper we demonstrate the presence of multimode resistive switching in Cu-O nanostructured thin film and the associated current conduction mechanism. The particle size distribution and elemental analysis are carried out using atomic force microscopy (AFM), crosssectional transmission electron microscopy (XTEM) and scanning tunnelling electron microscopy – high angle angular dark field (STEM-HADDF) imaging. In addition, x-ray photoelectron spectroscopy (XPS) is used to study the chemical composition of Cu-O nanostructures. Current-voltage characteristics measured using conductive atomic force microscopy (cAFM) reveal that filaments form in multistep processes, instead of the rapid one, indicating the presence of multimode resistive switching. The presence of multimode RS corroborates with defect-induced conduction mechanism across the Cu-O nanostructures. The present study should open up a new avenue to design a Cu-O-based advanced nanoscale memory device. 2. Experimental
3
Nanostructures of copper oxide, having a nominal thickness of 10 nm, was grown at room temperature (RT) on ultrasonically cleaned low resistive (ρ=1-510-3 Ω-cm) p-Si(100) wafers (area 11 cm2) by RF magnetron sputtering technique. The base pressure of the chamber (Excel Instruments, India) was 5×10-7 mbar and during the deposition of copper oxide film and the argon gas flow rate was maintained at 30 sccm, resulting in the working pressure of 5×10-3 mbar. The distance from the sample to the target was fixed at 80 mm and an RF power of 100 W was applied on the target. Microstructural analysis of copper oxide film was carried out by crosssectional transmission electron microscopy (XTEM) using a high-resolution transmission electron microscopy (HRTEM) ( FEI, Tecnai G2 F30-ST) operating at 300 kV equipped with a CCD camera (Gatan Orius). High-angle annular dark field (HAADF) scanning transmission electron microscopy (STEM) was employed using the same microscope, which was equipped with a scanning unit and a HAADF detector (Fischione, Model 3000). The compositional analysis was performed by energy dispersive X-ray spectroscopy (EDS) attachment (EDAX Inc.) on the Tecnai G2 F30. The compositional analysis of the Cu-O thin film was carried out using xray photoelectron spectroscopy (XPS) (VG Instruments). Further, the localized electrical transport phenomenon for forward and reverse voltage sweep of 10 V was investigated ex-situ by conductive atomic force microscopy (cAFM) (Asylum Research, MFP-3D). Conductive (Olympus, AC240TM) tip having ∼30 nm radius of curvature, ∼2 Nm−1 stiffness, and a resonance frequency of ∼70 kHz was used for electrical measurement. All recorded AFM images were analyzed using the WSxM software [10].
3. Results and discussion
4
Prior to exploring the RS behaviour in copper oxide, the sample is characterized by TEM and XPS. Figure 1(a) shows the cross-sectional TEM image of the copper oxide nanostructured film deposited on a Si substrate which clearly depicts the presence of uniformly distributed nanostructures, having an average size of 12±2 nm. Figure 1(b) shows a magnified view of the selected region marked on Fig. 1(a). A HRTEM image, collected from the marked region on Fig. 1(b) is presented in Fig. 1(c) which reveals the crystalline nature of Cu-O nanostructures. In fact, the d-spacing, calculated from this lattice image, matches well with the (111) plane of CuO phase. Likewise, HRTEM image collected from large randomly chosen places on the sample (images not shown) and d-spacings calculated from those matches closely with those of CuO and/or Cu2O phases. Thus, it becomes evident that both the phases (CuO and Cu2O) are present in the Cu-O nanostructured thin film. In addition, STEM-HAADF measurements confirm the presence of copper, oxygen, and silicon in the sample [Figs. 1(d)-(g)]. In addition, the morphology of Cu-O nanostructured thin film is studied by AFM measurement [Fig. 2(a)]. Here the inset depicts the grain size distribution estimated from this image where the average grain size turns out to be 12 nm. The compositional analyses are carried out by XPS. For instance, the high-resolution core level spectrum of Cu 2p is depicted in Fig. 2(b) [11] where the two peaks located at around 934.32 and 953.72 eV are attributed to the binding energy of Cu 2p3/2 and Cu 2p1/2, respectively, revealing an oxidation state of +2 [12]. Moreover, the two extra satellite peaks are also observed on a higher binding energy side, 942.0 and 962.2 eV for Cu 2p3/2 and Cu 2p1/2, respectively, implying the presence of an unfilled Cu 3d 9 shell and thus further confirms the existence of Cu2+ on the sample surface [13]. From the peak fitting of Cu 2p, we find two other small peaks, with binding energy of 932.6 and 952.6 eV, which are assigned to Cu 2p3/2 and Cu 2p1/2 in Cu2O, respectively [12,13]. The occurrence of a
5
satellite feature on the higher binding energy (943.2 eV) side of the Cu 2p main peak indicates the presence of CuO, because cupric oxide (Cu2+) has hole states in the Cu 3d band (Cu 3d9 configuration). This result indicates the deoxidization process and confirms the coexistence of CuO and Cu2O phases in the nanostructures (as discussed above). On the other hand, Fig. 2 (c) shows the XPS spectra corresponding to O 1s core level which can be deconvoluted into two Gaussian peaks, after Shirley background subtraction [14]. The broad and asymmetric nature of the peak can be ascribed to multiple co-ordinations of oxygen in the Cu-O nanostructures. The deconvoluted peaks at a binding energy of 529.0 and 531.8 eV are attributed to oxygen in Cu(OH)2 and Cu2O, respectively [15]. Figure 3(a) shows the local I-V characteristics of nanostructures, using cAFM measurements, obtained by sweeping a DC bias according to 0 V → 10 V → 0 V → -10 V → 0 V with a ramp rate of 0.2 V s-1 at a fixed tip location on a randomly chosen topographical places. It is noted that nanostructures exhibit bipolar switching behavior: it sets at positive bias and resets at a negative one corresponding to the current compliance of 20 nA. The I-V curve, presented in Fig. 3(a) is separately analyzed for different branches: at positive voltages, the current is very low and remains constant (path #1) until the voltage reaches around +7.5 V from where the current increases suddenly (path #2) and the corresponding voltage is identified as the ‘set’ voltage (Vset). It is interesting to note that during the set processes highly reproducible multiple current jumps are detected. This particular feature is not generic and is observed at wide range of topographical locations. Beyond 7.5 V, the current value reaches the compliance limit (20 nA) of the cAFM setup and hence, shows a straight line (path #3). On the other hand, during the decreasing cycle of voltage, current remains saturated till far below Vset (7.5 V). At much lower voltages (<3 V), current decreases almost linearly, indicating the Ohmic nature of the tip-
6
sample contact (path #4). A small deviation from the ohmic behaviour is due to the presence of bottom Si substrate. It is worth mentioning here that I-V characteristics are recorded from a large number of randomly chosen points on the sample to observe that behaviour of the curves remains nearly similar. A pronounced change in resistance is also observed for both increasing and decreasing voltages, corresponding to a tip bias (Vtip) of 5 V which is known as high resistance state (HRS) and low resistance state (LRS), respectively. Subsequently, an opposite ‘reset’ process is also observed when sweeping the voltage from 0 to -6 V (Vreset, path #5), as is evidenced by a two-step switching from LRS to HRS [Fig. 3(a)]. Figure 3(b) shows the forward bias set I–V curves in the forming voltage region in semilog scale. For a better clarity, a zoomed part of Fig. 3(b) is also depicted in Fig. 3(c). It is interesting to note that during the SET process, current does not increase suddenly but instead it goes through steps. This indicates that a particular process is not responsible for the filament formation and instead multiple-mode processes take place during the same. The exact multimode mechanism during the set process is still under investigation, yet the possible reason behind this may be a sudden increase in the current which can generate heat within the nanostructures and in turn induce defect injection, resulting an improvement in the filament size [Fig. 4(a)] [16,17,18,19]. In fact, the size of the filament can change due to defect injection/extraction during the set process [Fig. 4(b)]. The observed effect is quite significant since it suggests that the defects are collected by applied voltage. On the other hand, the reset process is attributed to the thermal dissolution of the metal filament due to Joule-heating-assisted oxidation followed by the diffusion of metal ions under the concentration gradient and the applied electric field (the ON and OFF states are schematically illustrated in the insets) [Fig. 4(c)]. The migration of oxygen vacancies can be described by two mechanisms: (i) drift and (ii) diffusion [19,20]. The driving
7
forces for drift and diffusion are the gradients of the electrostatic and electrochemical potentials, respectively. The diffusion coefficient and mobility are the most important parameters essential for describing the voltage-dependent migration behaviour of charged particles including oxygen ions and corresponding defects. Such pathways are around 5–30 nm in diameter, suggesting that devices may be scaled down to the nanometer dimensions to achieve very high levels of integration.
4. Conclusions In conclusion, we have observed multiple level SET process in the RS phenomenon in Cu-O nanostructures. AFM, XTEM, STEM-HADDF, and XPS studies are employed to evaluate sample morphology, extract particle size distribution, study the microstructure, and find out the atomic compositions. It is observed that crystalline Cu-O nanostructured thin film grown at RT shows the coexistence of both CuO and Cu2O phases. cAFM measurements reveal that filament formation takes place in multistep processes, instead of a rapid one, confirming the multimode resistive switching. The observed results are explained in terms of thermally-induced defect injection and migration and in turn improvement in the filament size. The model suggests that the optimization of the conductive filament active region is key for the future nanoscale resistive switching devices. The present results will be a benchmark to design efficient resistive Cu-O nanostructure-based memory devices.
Acknowledgements The authors thank to Professor Shikha Varma and Mr. Santosh Kumar Choudhury from Institute of Physics, Bhubaneswar for the XPS measurements.
8
References [1] K. M. Kim, D. S. Jeong, and C. S. Hwang, Nanotechnology 22 (2011) 254002. [2] R. Waser, R. Dittmann, G. Staikov, and K. Szot, Adv. Mater. 21 (2009) 2632. [3] Mario Lanza, Materials 7 (2014) 2155. [4] E. Meyer, H. J. Hug, R. Bennewitz Scanning Probe Microscopy. The Lab on a Tip SpringerVerlag, Heidelberg, Germany (2004). [5] M. Kumar and T. Som Nanotechnology 26 (2015) 345702. [6] K. Park and J.-S. Lee Nanotechnology 27 (2016) 125203. [7] A. Chen, S. Haddad, Y. C. Wu, Z. Lan, T. N. Fang, and S. Kaza, Appl. Phys. Lett. 91 (2007) 123517. [8] S. Y. Wang, C. W. Huang, D. Y. Lee, T. Y. Tseng, and T. C. Chang, J. Appl. Phys. 108 (2010) 114110. [9] J. W. Han and M. Meyyappan, AIP Adv. 1 (2011) 032162. [10] http://www.nanotec.es/products/wsxm/ [11] J. P. Espinos, J. Morales, A. Barranco, A. Caballero, J. P. Holgado, and A. R. GonzalezElipe, J. Phys. Chem. B 106 (2002) 6921. [12] C.-K. Wu, M. Yin, S. O’Brien, and J. T. Koberstein, Chem. Mater. 18 (2006) 6054. [13] J. Morales, J. P. Espinos, A. Caballero, A. R. Gonzalez-Elipe, J. A. Mejias, J. Phys Chem B. 109 (2005) 7758.
9
[14] Y. Wang, S. Lany, J. Ghanbaja, Y. Fagot-Revurat, Y. P. Chen, F. Soldera, D. Horwat, F. Mucklich, and J. F. Pierson, Phys. Rev. B 94 (2016) 245418. [15] J. Morales, L. Sanchez, F. Martın, J. R. Ramos-Barrado, M. Sanchez, Thin Solid Films 474 (2005) 133. [16] H. D. Lee, B. Magyari-Köpe, and Y. Nishi, Phys. Rev. B 81 (2010) 193202. [17] S. Nigo, M. Kubota, Y. Harada, T. Hirayama, S. Kato, H. Kitazawa, and G. Kido, J. Appl. Phys. 112 (2012) 033711. [18] U. Celano, G. Giammaria, L. Goux, A. Belmonte, M. Jurczaka and W. Vandervorsta, Nanoscale, 8 (2016) 13915. [19] K.-C. Kwon, M.-J. Song, K.-H. Kwon, H.-V. Jeoung, D-W. Kim, G-S. Lee, J.-P. Hong and J.-G. Park, J. Mater. Chem. C, 3 (2015) 9540. [20] Y. Hou, U. Celano, L. Goux, L. Liu, R. Degraeve, Y. Cheng, J. Kang, M. Jurczak, and W. Vandervorst, Appl. Phys. Letts. 109 (2016) 023508. [21] Yuchao Yang and Wei Lu, Nanoscale, 5 (2013) 10076.
10
List of figures FIG. 1. (Color online) (a) The XTEM image of Cu-O nanostructures. (b) Shows the magnified view of the marked region on (a). (c) Shows a HRTEM image obtained from the marked region on (b). STEM-HAADF elemental map of Cu, O and Si, and total are shown in (d)-(g), respectively.
FIG. 2. (Color online) (a) AFM image of the as-deposited Cu-O thin film on a Si substrate. The inset shows the particle size distribution (b) and (c) Depict Cu 2p and O 1s XPS spectra, respectively obtained from the Cu-O nanostructured film.
FIG. 3. (Color online) (a) Local I-V characteristics performed at various points, whereas the semi-log of the forward bias for the same is depicted in (b). For a clarity, zoomed part of the (b) shows in (c).
FIG. 4. (Color online) Schematic diagrams of filament formation and its growth are depicted in (a) and (b), respectively. (c) reset process, presenting the filament breakdown.
11
Figure 1 Kumar et al.
12
Figure 2 Kumar et al.
13
Figure 3 Kumar et al.
14
Figure 4 Kumar et al.
15
Highlights
1. Depicting nanoscale multimode resistive switching (RS) in nanostructured copper oxide thin film 2. Utilization of conductive atomic force microscopy for studying this nanoscale behavior 3. XTEM and STEM-HADDF imaging confirms the formation of Cu-O nanostructures 4. RS corroborates with defect-induced conduction mechanism
16
Graphical abstract
17