Dipartimento di Fisica Corso di Dottorato di Ricerca in Fisica, Astrofisica e Fisica Applicata Ciclo XXVIII Oxide-based memristive devices by block copolymer self-assembly Settore Scientifico Disciplinare FIS/03 Supervisore: Prof. Alberto Pullia Co-supervisore: Dott.ssa Sabina Spiga Tesi di dottorato di: Jacopo Frascaroli Coordinatore: Prof. Marco Bersanelli Anno Accademico 2014 – 2015 Commission of the final examination: Prof. Thomas Mikolajick (NaMLab and TU Dresden) Prof. Alberto Pullia (UNIMI) Prof. Giuseppe Bertuccio (POLIMI) Dr. Sabina Spiga (MDM, IMM–CNR) Final examination: January 25, 2015 Universita` degli Studi di Milano, Dipartimento di Fisica, Milano, Italy The Ph.D acvitity was carried out at MDM laboratory: Institute for Microelectronics and Microsistems, section of Agrate Brianza (MB), belonging to the National Research Council of Italy. CNR - IMM Laboratorio MDM Cover illustration: Moire´ pattern produced by superimposing a regular grid over a pattern created by self- assembled block copolymer lamellae. The appearance of the Moire´ pattern denotes the regularity of block copolymer nanostructures, with a constant spacing among features. This pattern can also be used to visually distinguish among different domains in which lamellae have parallel orientation. Moire´ patterns are routinely observed during scan- ning electron microscopy analysis due to the horizontal scan lines and can also be visu- alized in monitors whenever the pixel grid spacing approaches the periodicity of block copolymer features. Abstract OXIDE-BASED MEMRISTIVE DEVICES BY BLOCK COPOLYMER SELF-ASSEMBLY Oxide-based memristive systems represent today an emerging class of devices with a significant potential in memory, logic, and neuromorphic circuit applications. These devices have a simple capacitor structure and promise superior scalability together with favorable memory performances. This thesis presents a study of resistive switching phe- nomena in HfOx-based nanoscale memristive devices, with focus on material properties and development of bottom-up approaches for the fabrication of structures with dimen- sion down to the nanoscale. One of the main issues for practical applications regarding device variability is first assessed by doping hafnium oxide films with different concentrations of aluminum at- oms. Testing devices are analyzed by physico-chemical and electrical techniques in or- der to define the effect of oxide doping on the device properties. In the following part of the thesis, the scalability limit is explored in very high density arrays of nanodevices produced exploiting a lithographic approach based on the bottom-up self-assembly of block copolymer templates. This technique allows a tight control over the size and den- sity of the defined features, and the possibilities offered by block copolymer patterning are here discussed. Electrical measurements of the nanodevices are performed through conductive atomic force microscopy. The device variability is examined and related to the inherent oxide non-homogeneity at the nanoscale, while a non-volatile switching of the resistance of the nanodevices is demonstrated. Further, this analysis draws the at- tention to a crosstalk phenomenon occurring at the nanoscale in a continuous thin film geometry. This result suggests to select different system configurations. A promising technique based on selective reactions with one copolymer block is finally discussed which allows the direct production of oxide patterns from block copolymer templates avoiding a pattern transfer process. In conclusion, the results reported in this thesis highlight the high scalability potential of oxide-based memristive devices, providing a missing piece of information for the understanding and practical development of very high density arrays. iii Contents List of Figures ix List of Tables xiii Introduction xv Motivation and objective of the activity xv Brief introduction to the chapters xviii 1 Resistive switching memories 1 1.1 Introduction1 1.2 Resistive switching memories: switching behavior3 1.2.1 Unipolar or non-polar4 1.2.2 Bipolar4 1.2.3 Complementary switching5 1.3 Resistive switching memories: mechanisms and physics5 1.3.1 Thermochemical mechanism5 1.3.2 Electrochemical metallization mechanism6 1.3.3 Valence change mechanism8 1.4 Resistive switching in hafnium oxide9 1.4.1 Evidences of filamentary switching 10 1.4.2 Role of the metal electrodes 12 1.4.3 Switching variability 13 1.4.4 Oxide doping 15 1.5 Applications and integration schemes 17 1.5.1 Resistive switching device as memristor: the fundamental circuit element 19 2 Bottom-up fabrication approaches for nanoscale resistive switching devices 27 2.1 Introduction 27 2.2 Template assisted fabrication of oxide nanostructures 28 2.2.1 Anodic alumina oxide (AAO) membranes 28 2.2.2 Nanosphere lithography (NSL) 30 2.2.3 Block copolymer templated self-assembly (TSA) 31 2.2.4 Nanowire growth by bottom-up techniques 32 v vi Contents 2.2.5 Synergistic combination of top-down and bottom-up approaches 34 2.3 Resistive switching in oxide nanostructures 34 2.3.1 Resistive switching in multiple nanostructures 35 2.3.2 Resistive switching in a single nanostructure 36 2.3.3 Resistive switching in core–shell nanowires 40 3 HfO2-based resistive switching devices 47 3.1 Introduction 47 3.2 Deposition and characterization of the materials 47 3.2.1 XPS characterization of the oxide film and of its interfaces 49 3.3 Electrical characterization 55 3.3.1 Bipolar voltage sweeps 56 3.3.2 Device area dependence 59 3.3.3 Role of the metal–oxide interface on the switching behavior 60 3.4 Effect of Al doping on the DC switching behavior 62 3.4.1 Variability and endurance properties 63 3.4.2 Retention measurements 66 3.5 Current fluctuations and random telegraph noise 72 4 Block copolymer self-assembly for lithographic applications 83 4.1 Introduction 83 4.2 Block copolymer self-assembly 84 4.2.1 Applications 89 4.3 Periodic array of metal nanoparticles on the HfO2 surface 91 4.3.1 HfO2 Surface neutralization 92 4.3.2 Block copolymer template formation 95 4.3.3 Pattern transfer process 97 4.3.4 Highly scaled, high density array of patterned top electrodes for resistive switching memories 102 5 Characterization of high density arrays of resistive switching devices 111 5.1 Introduction 111 5.2 Nanoscale conduction variability in HfO2 thin films 112 5.3 Characterization of nanoscale RRAM arrays fabricated by block copoly- mer templates 116 5.3.1 Initial state characterization of a large number of nanoscale devices 116 5.3.2 Resistive switching at the nanoscale 119 5.3.3 Crosstalk observation between distinct nanodevices 121 5.3.4 Concluding remarks 126 6 Future outlook: oxide RRAM patterned by sequential infiltration synthesis 131 6.1 Introduction 131 6.2 PS-b-PMMA block copolymer templates 133 6.3 Sequential infiltration synthesis in PS-b-PMMA block copolymer 134 6.3.1 Effect of temperature 135 6.3.2 Effect of precursor pressures 137 6.3.3 Study as a function of the number of SIS cycles 138 6.3.4 XPS analysis of the nanostructured alumina 141 6.3.5 Sequential infiltration in random copolymer 143 Contents vii 6.3.6 Cylindrical structures 143 6.4 Perspectives 144 6.5 Conclusions of the thesis 146 A Experimental techniques 151 A.1 Deposition methods 151 A.1.1 Sputter deposition 151 A.1.2 electron beam evaporation 152 A.1.3 Atomic Layer Deposition 153 A.2 Rapid thermal processing 154 A.3 Top electrodes patterning by UV lithography 156 A.4 X-Ray Photoemission Spectroscopy 156 A.5 Scanning electron microscopy 157 A.6 Atomic force microscopy 158 List of Publications 164 Acknowledgments 166 List of Figures 1.1 Filamentary switching.2 1.2 Switching schematic.3 1.3 I–V curves and thermal profile in a TCM cell.6 1.4 ECM cell characteristics and processes.7 1.5 VCM cell characteristics and processes.8 1.6 Evidences of filamentary CF in HfO2-based memory cell by C–AFM-assisted 3D tomography. 11 1.7 Evidences of filamentary CF in HfO2-based memory cell by STEM analysis. 12 1.8 Monte Carlo simulation of the switching variability by random variation of the energy barrier for vacancies injection in the filament. 14 1.9 Monte Carlo simulation of random filament formation in a 10 × 10 nm2 HfO2 cell during forming and reset processes. 14 1.10 HfO2 doping: supercell, wave function contour plot for the gap-state in correspondence of a dopant, and calculated formation energies for oxygen vacancies. 15 1.11 Effects of doping: reduction of the forming voltage and associated re- duced ON/OFF ratio; reduction of the energy barrier for migration of the oxygen species. 16 1.12 Improvement of the resistance and switching voltage distributions upon Al doping. 16 1.13 Crossbar array architectures. 17 1.14 Illustrative applications of resistive switching devices. 18 1.15 Conceptual symmetries of passive circuit elements. 19 2.1 AAO template formation and electrodeposition 29 2.2 Schematic illustration of Ni nanowire synthesis by electrodeposition through AAO membranes 29 2.3 One monolayer colloidal mask for nanosphere lithography 30 2.4 Examples of self-aligned patterns formed by block copolymer directed self-assebly. 32 2.5 VLS growth mechanism schematic and ZnO nanorods grown using Au droplets 32 2.6 Example of fully bottom-up process for the growth of organized nanowires by VLS 33 ix x LIST OF FIGURES 2.7 Resistive switching in multiple nanowires 35 2.8 One diode – one resistor fabrication procedure using NSL 36 2.9 Cobalt oxide nanowire grown by VLS and contacted in a multi electrode configuration. 37 2.10 Single ZnO NW contacted by two Ti electrodes and relative multi-state switching operations 38 2.11 Schematic of the experimental set-up used to probe individual NWs by C–AFM 39 2.12 NiO heterostructured nanowires and relative I–V characteristics.