[论文解读] An Investigation into the $HfO_2/Si$ Interface: Materials Science Challenges and their Effects on MOSFET Device Performance

Aditya Muralidharan|arXiv (Cornell University)|Jun 20, 2022

Semiconductor materials and devices被引用 4

一句话总结

本文研究了MOSFET中HfO₂/Si界面处界面SiO₂层形成的热力学与动力学机制，确定沿晶界扩散的氧以及SiO物种为主要原因。研究证明，通过控制氮气退火和优化原子层沉积（ALD）工艺，可抑制中间层生长，从而降低等效氧化层厚度（EOT）和漏电流，同时保持高迁移率与阈值电压稳定性。

ABSTRACT

Since the 1960's when Gordon Moore proposed that the transistor density in our electronic devices should double every two years while the cost is halved, the semiconductor industry has taken this statement to heart. Over the last few decades, no other industry has seen growth even comparably close to that experienced by the semiconductors industry. This has all been made possible by the unbroken string of ingenious breakthroughs by brilliant minds that have been working tirelessly to shrink down transistors. The latest of which is the use of high-k dielectrics and a return to metal gates combined with 3D-transistor architectures. This has been the enabling technology for the transition from the 90 nm node to the 45 nm node, allowing us to shrink our transistors further without losing additional gate control. The fundamental reason for using a high-k gate dielectric compared to SiO2 is that shrinking our gate oxide further, which is already at a few angstroms, is no longer a feasible option to gain additional gate control. High-k dielectric overcome this by exploiting the fundamental physics of capacitors and the materials science of dielectrics to provide a viable option to increase gate control without the need for successively thinner gate oxides. Hafnium Oxide is the most studied and popular of such materials. Its high dielectric constant ~16-25 and interface stability with silicon at operating temperatures make it an ideal candidate for use in current CMOS technology. Despite its deceivingly simple appearance, the processes involved in the fabrication of such high-k HfO2/Si interfaces are full of process subtleties and nuances. In this term paper we hope to explore the physics and materials science of these high-k HfO2/Si interfaces, discussing the challenges and ways to overcome them when it comes to its actual fabrication, and how this ultimately affects our device performance.

研究动机与目标

• 理解在高温处理过程中HfO₂/Si界面处SiO₂中间层形成的热力学与动力学驱动力。
• 识别导致EOT劣化及高k MOSFET器件性能下降的界面层生长的根本原因。
• 评估退火气氛、氧气流量以及沉积方法（PEALD与热ALD）等工艺参数对抑制中间层形成的效能。
• 关联HfO₂基MOSFET中中间层厚度与缺陷密度，以及栅极漏电流和阈值电压漂移的关系。
• 为优化HfO₂/Si界面提供材料科学框架，以支持CMOS技术在45nm节点之后的持续微缩。

提出的方法

• 利用吉布斯自由能（ΔG₀ = 227.9 kJ/mol）进行热力学分析，评估HfO₂/Si体系的稳定性，排除热力学驱动力作为中间层形成的主要原因。
• 分析相图以评估界面在约1700 K以内的稳定性，表明仅靠稳定性无法解释中间层的生长。
• 研究氧扩散路径，特别是多晶HfO₂中沿晶界的扩散，因正电性氧空位促进离子迁移。
• 评估在氧化条件下，界面处SiO物种扩散对中间层形成的贡献作用。
• 比较通过等离子体增强ALD（PEALD）与热ALD沉积的HfO₂薄膜的漏电流与开关特性，将非晶结构归因于更低的漏电流。
• 评估氮气气氛退火以及沉积过程中氧气流量持续时间对中间层厚度与缺陷钝化的影响。

实验结果

研究问题

• RQ1 在高温处理过程中，HfO₂/Si界面处SiO₂中间层形成的热力学与动力学主驱动力是什么？
• RQ2 氧沿多晶HfO₂中晶界扩散以及SiO物种迁移如何促进中间层的生长？
• RQ3 氮气气氛退火或沉积过程中控制氧气流量在多大程度上可减少中间层厚度与漏电流？
• RQ4 HfO₂的晶体结构（非晶态与多晶态）如何影响栅极漏电流与缺陷密度？
• RQ5 中间层形成（导致EOT增加）与对载流子迁移率和阈值电压稳定性带来的有益影响之间存在何种权衡？

主要发现

• 在多晶HfO₂中，由正电性氧空位促进的氧沿晶界扩散是中间层形成的主要机制，而非热力学不稳定性。
• 在氧化性退火条件下，由于氧扩散增强，中间层厚度增加，直接与更高的EOT和更大的栅极漏电流相关。
• 氮气气氛退火显著减少中间层形成与漏电流，可能由于HfO₂及其界面层部分氮化，从而抑制氧的迁移能力。
• PEALD沉积的HfO₂薄膜漏电流特性优于热ALD薄膜，归因于其非晶结构可最小化缺陷传输路径。
• 在沉积过程中控制氧气流量可消除晶界处的氧空位，这些空位原本会作为电子导电通道。
• 退火循环可使栅极漏电流降低数个数量级，主要通过缺陷消除与捕获态钝化实现，其中氮气气氛退火表现出显著潜力。