Electronic switching in phase-change materials Von der Fakult¨atf¨urMathematik, Informatik und Naturwissenschaften der RWTH Aachen University zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften genehmigte Dissertation vorgelegt von Diplom-Physiker Gunnar Bruns aus Telgte Berichter: Univ.-Prof. Dr.rer.nat. Matthias Wuttig Univ.-Prof. Dr.-Ing. Rainer Waser Tag der m¨undlichenPr¨ufung:31.01.2012 Diese Dissertation ist auf den Internetseiten der Hochschulbibliothek online verf¨ugbar. Acknowledgements An interdisciplinary work like this de- have been advisors in all fields of physics. mands for perfect interplay between dif- Regarding phase-change switching effects ferent affiliations and people with different and resistance drift Daniel Krebs was an in- aims and different ways of thinking and dispensable discussion partner. working. I am grateful to all those people A close and fruitful teamwork with my which have enabled the fast and successfully three diploma students, Martin Wimmer, progressing projects of my work. Karsten Fleck, and R¨udigerSchmidt, has I acknowledge the opportunity to con- lead to exciting experiments and results tribute to the fascinating field of phase- which can be found all over this work. change material research which has been Stephan Hermes skills and his commitment made possible by my PhD-advisor Matthias maintains our equipment, and therefore, se- Wuttig. I would like to thank Rainer Waser, cures a failure free work environment. The the second advisor of this work, for the fruit- precise and fast work of our team in the ful projects with his department. mechanical and electronic workshops, espe- The custom made setups which have been cially Axel Gross, Franz Neus, and Gert build in this work have been created to- Kirchhoff, were indispensable for the relia- gether with Carl Schlockermann in a close bility of the custom made setups. Ralf De- and complementary work. Philipp Merkel- temple, Sarah Schlenter, and Josefine Elbert bach has developed and performed most of enabled a simple and productive manage- the lithographic processes. He and Peter ment of fundings, equipment acquisitions, Zalden were great office mates, and they and all other aspects of administrations. were the first address to discuss new results Special topics and aspects of this thesis and to solve problems. have been discussed with Janika Boltz, Peter The Qimonda project allowed a close con- Jost, Andreas Kaldenbach, Michael Klein, nection to Jan Boris Philipp, Thomas Happ, Anja K¨onig, Stephan Kremers, Dominic and Michael Kund. This project was a per- Lencer, Jennifer Luckas, Hanno Volker, and fect example for an industry-university joint Michael Woda. venture and was a driving force towards the results of this thesis. I'd like to thank my family and all my friends Martin Salinga and Daniel Wamwangi for their support. III Abstract Phase-change materials (PCM) possess regarding the memory switching speed in a unique property contrast between their phase-change memory will be presented in crystalline and amorphous phases. Differ- this work, and reveal that the phase tran- ences in resistivity of some orders of magni- sitions can be accomplished within a few tude can be observed between both phases. nanoseconds. This demonstrates the poten- Also, the reflectivity shows a remarkable tial of PCMs to compete with DRAM in contrast, and it allows the application of terms of speed. thin phase-change layers in optical media, The second topic in this work are tran- such as CD, DVD and Bluray disk, to enable sient phenomena, like threshold switching, rewritable storage of information. Although which occur when PCMs are treated with both phases are stable for decades at room electrical pulses. Threshold switching de- temperature, it is possible to switch between scribes a sudden decrease of the material's the phases in nanoseconds at elevated tem- resistivity. This effect can be observed in peratures. This striking combination of sta- amorphous PCMs, if the applied electrical bility and rapid transition, together with the field exceeds a threshold value. In this pronounced resistivity contrast, make PCMs work, results will be presented which de- one of the most promising candidates for scribe both the resistance drop during the future, non-volatile, electronic memory. In threshold switching and the life time of this this work, three physical aspects concern- high conductive state. While present pub- ing such a memory have been investigated lications use a characteristic field strength using custom made setups to cope with the for each PCM to describe the occurrence challenges of sub-nanosecond timescales and of the threshold switch, the results of this resistances of hundreds of Gigaohms. work suggest the definition of a field depen- Speed limitations of write and rewrite op- dent delay time, which predicts the sudden erations are a crucial topic, if PCMs shall be change of conductivity. able to compete with the established con- Besides the extraordinary behavior of cepts in electronic devices. Two classes of disordered semiconductors at high electric storage devices are used in modern computer fields, there is a further effect at low fields, systems, so far. On the one hand, there is which has a tremendous influence on phase- the fast, but volatile memory close to the change memory applications: the resistance processor unit, like the dynamic (DRAM) drift. This effect describes the time depen- and the static (SRAM) random access mem- dent increase of the resistivity of amorphous ory. On the other hand, there are slower, but PCMs. In this work, experimental data will non-volatile storages, like hard disk drive, be presented which demonstrate the similar- flash, and optical media. In terms of speed, ity of this effect in both unstructured films there is a gap of several orders of mag- and memory devices. These data have been nitude between memory and storage con- used to modify an existing model which ex- cepts. Phase-change memory could close plains the drift's origin. The reported de- this gap and establish a new storage class pendency of the drift behavior on the activa- memory, and maybe, even allow to build tion energy for conduction can be confirmed a non-volatile memory device, which could and a new aspect of this dependency will be replace the volatile DRAM. Investigations presented. IV Kurzfassung Ubersetzung¨ des englischen Originaltitels: Elektronisches Schalten in Phasenwech- selmaterialien. Phasenwechselmaterialien (PCM) be- gibt es die schnellen, aber fl¨uchtigen Ar- sitzen einen einzigartigen Kontrast physi- beitsspeicher (memory), wie DRAM (dy- kalischer Eigenschaften zwischen ihrer namic random access memory) und SRAM kristallinen und amorphen Phase. So un- (static random access memory). Zum terscheidet sich der spezifische Widerstand anderen sind da die langsameren, aber beider Phasen in manchen Materialien um nicht fl¨uchtigen Speicher (storage), wie Fest- mehrere Gr¨oßenordnungen. Ebenso gibt platte, Flash und optische Medien. Unter es einen beachtlichen Kontrast im Reflexi- dem Blickwinkel der Geschwindigkeit klafft onsverm¨ogen, der bereits industrielle An- eine L¨ucke von mehrerer Gr¨oßenordnun- wendung gefunden hat: in optischen Me- gen zwischen den Konzepten Arbeitsspe- dien wie CD, DVD und Bluray Disks icher und Langzeitspeicher. Phasenwechsel- erm¨oglichen d¨unne Schichten aus PCM Speicher k¨onnten diese L¨ucke schließen eine mehrfach wiederbeschreibbare Daten- und eine neue Speicherklasse (storage speicherung. Und obwohl beide Phasen class memory) schaffen und eventuell an Raumtemperatur f¨ur Jahrzehnte sta- sogar einen nicht fl¨uchtigen Arbeitsspe- bil sind, ist es bei h¨oherenTemperaturen icher erm¨oglichen, der den fl¨uchtigen DRAM m¨oglich, innerhalb von Nanosekunden zwis- ersetzt. In dieser Arbeit werden Un- chen den Phasen zu wechseln. Diese er- tersuchungen der Schaltgeschwindigkeiten staunliche Kombination von Stabilit¨atund in Phasenwechsel-Speichern pr¨asentiert, die schnellen Uberg¨angen,gemeinsam¨ mit dem belegen, dass der Phasen¨ubergang inner- ausgepr¨agten Eigenschaftskontrast, macht halb weniger Nanosekunden stattfinden PCMs zu einem vielversprechenden Kandi- kann. Dies demonstriert das Potential der daten f¨urzuk¨unftige,nicht fl¨uchtige, elek- PCMs unter dem Gesichtspunkt der Schalt- tronische Speicher. In dieser Arbeit wur- geschwindigkeit mit dem DRAM mithalten den drei physikalische Aspekte solcher Spe- zu k¨onnen. icher untersucht, wof¨urspeziell angefertigte Das zweite Thema dieser Arbeit sind die Messpl¨atzegeschaffen wurden, um die Her- Ubergangsph¨anomene,die¨ auftreten, wenn ausforderungen zu meistern, die bei Mes- PCMs hohen elektrischen Feldern ausge- sungen auf der Picosekunden-Skala und mit setzt werden. Das Threshold switch- " Widerst¨anden von mehreren hundert Gi- ing\ beschreibt die pl¨otzliche Abnahme gaohm auftreten. des spezifischen Widerstandes. Dieser Ef- Die Limitierung der Geschwindigkeit von fekt kann in amorphen PCMs beobachtet Schreib- und L¨osch-Operationen ist ein werden, sobald die angelegten Feldst¨arken wesentliches Thema bei der Realisierung auf einen Schwellenwert ¨uberschreiten. In dieser PCMs basierender elektronischer Speicher, Arbeit werden Ergebnisse pr¨asentiert, die die mit den bereits etablierten Konzepten sowohl den Einbruch des Widerstandes mit konkurrieren sollen. Zur Zeit werden dem Threshold switching\ beschreiben, als " zwei Klassen zur Speicherung elektronischer auch die Lebenszeit des damit verbun- Daten in Computern genutzt. Zum einen denen angeregten Zustandes