Growth and Characterization of Ferroelectric Lanthanum-Doped Hafnia

Abstract

Hafnia-based ferroelectrics show great promise as future nonvolatile memory devices, however, their issues regarding device inconsistency across their lifetime, coupled with the relatively short total lifetime, makes these devices only theoretical as of now. In this thesis, an ALD deposition recipe for lanthanum oxide deposition was created. Using this recipe, lanthanum-doped hafnia thin films were manufactured and characterized. Although several issues emerged which limited the ALD-step of the process, ferroelectric devices were created and characterized. The devices have a coercive field between 0.9 and 1.2 MV/cm with remanent polarizations of up to 20 µC/cm2. Endurance measurements showed no sign of fatigue even after 10 million cycles - indicating a relatively long lifetime of the devices. This long lifetime is accompanied by a very long wake-up, which could be due to the issues with the ALD recipe. A comparison between Rapid Thermal Annealing (RTA) and Flash Lamp Annealing (FLA) as the selected annealing method was made. Although the results indicate no ferroelectricity in the samples annealed in the FLA, no conclusion can be drawn as to whether this is due to the FLA process or underlying problems with the oxide on the samples.

Popular Abstract

Memory is an oftentimes hidden aspect of computing, and has not until very recently been a limiting factor in what types of tasks can be performed on a computer. However, the nature of memory today causes it to drain a lot of energy, an issue which would be very beneficial for both computer performance and energy consumption. A typical computer has some long-term memory, such as a hard-disk drive, or in more recent years, a solid state drive. These memories are nonvolatile, meaning that once written to, they retain their information for decades years without losing it. The downside with these memories is that they are so slow that the brains of the computer, the processor, could have done millions of calculations in the time it takes the memory to change a bit of information [1]. This necessitates some high-speed memory which is in direct communication with the processor. Typically, this high-speed memory consists primarily of Dynamic Random Access Memory, or DRAM, which allows read- and write operations to occur in tens of nanoseconds. The problem with this memory is that it leaks its memory state with time. Imagine DRAM like a reservoir of water connected with a tap. If you open the tap, you can find out if there is water or not in the reservoir, but a hole in the bottom makes the water leak out over time. This problem results in the need to constantly refresh DRAM. This is done once every 64 milliseconds, and is a large reason for power consumption in this high-speed memory [2]. The ”holy grail” of memory technology is thus some memory device which is quick, but retains its information over long periods of time, all while being compact enough for gigabytes of data storage. There are devices which could work for these purposes, but they face several issues which prevent them from reaching markets. One type of these memory devices are based on so-called ferroelectric films. These memories are based on materials with a specific crystalline structure which shifts its properties in measurable ways when subjected to electric signals. These shifts are permanent and quick, solving both of these problems. While these memories are developed to the point where they have been implemented in actual market-viable products, they faced one primary issue. The devices are too large, and needed shrinking in order to become viable for large-scale integration. Recent developments in this field have discovered alternative materials which can become ferroelectric, while allowing for reasonable scaling. These techniques have to be refined and research is currently investigating ways of creating these memory devices while solving some of the issues they are facing. This thesis is looking in to just that, and is focused on developing and characterizing ferroelectric thin films, which may in the future find a home in your computer.
[1] Wong, H., Salahuddin, S. Memory leads the way to better computing. Nature Nanotech 10, 191194 (2015). [2] Dynamic RAM, DRAM Memory Technology, www.electronics-notes.com

• author: Havir, Harald
• supervisor: Mattias Borg
• organization: Solid State Physics; Department of Physics
• course: FYSM30 20201
• year: 2020
• type: H2 - Master's Degree (Two Years)
• subject: Science General
• keywords: Ferroelectricity, Lanthanum-doped Hafnia, ALD
• language: English
• id: 9016131
• date added to LUP: 2020-06-15 09:05:41
• date last changed: 2020-06-15 09:05:41