Introductory Invited paperProgress in device isolation technology
Introduction
Without device isolation technology, the realization of integrated circuits as we know today, like processors, dynamic or non-volatile memories and system-on-chip applications would not have been possible. For several decades, local oxidation of silicon (LOCOS) [1] was the ultimate isolation technique used in MOS and bipolar technologies to ensure undisturbed operation of thousands of integrated active devices. However, as technologies were approaching the 0.25 μm design rule, the limitations of LOCOS isolation became more and more apparent. Besides the substantial lateral active-area loss due to field oxide encroachment, the non-planarity between isolation oxide and active area also caused increased difficulties for lithography and etching. In order to overcome this bottleneck, shallow-trench-isolation (STI) [2], [3] was introduced. STI is scalable to extremely high packing densities because of the abrupt transition between active device and isolation regions and it provides the desired planar surface topology. However, these advantages come not for free. STI processing is much more complex than previous LOCOS technology and demands for a much tighter process control. Besides being more expensive, STI adversely affects device characteristics. In particular the enhanced sub-threshold leakage (parasitic corner effect) [4] and the narrow width effect [5] are among the major drawbacks. Moreover, the perimeter component will become more important in future technology generations since the area-to-perimeter ratio tends to decrease in further downscaled devices.
This contribution briefly reviews the current status on device isolation technology. Starting with conventional STI, the challenges introduced by this approach are outlined. Based on this discussion, the process concept of the recently developed extended trench isolation gate technology (EXTIGATE) [6], [7], [8], [9], [10], [11], [12], [13] is presented. It will be shown that EXTIGATE not only provides a relief from the drawbacks of conventional STI but also offers promising alternatives for front-end process integration, device optimization and specific applications as well.
Section snippets
Challenges of conventional STI-CMOS
Conventional STI suffers from problems inherently related to the abrupt transition between active device and isolation regions, namely the trench corner. Since CMOS processing starts with the definition of active area and isolation regions, subsequent cleaning and etching sequences prior to gate oxidation often create under cuts in the isolation oxide which cause gate-wrap-around [4] (Fig. 1, left inset). In combination with oxide thinning and dopant loss at the side wall, the crowding of gate
Process concept
The EXTIGATE structure [6], [7], [8], [9], [10], [11], [12], [13] basically results from the fact that the essential part of the MOS gate is fabricated prior to device isolation on the bare wafer, as shown in Fig. 3. After wafer clean, the process starts with gate oxidation followed by bottom poly-Si and nitride deposition. For ultra-thin oxides this first step is favorably done by in situ cluster tool processing [11], [12], as discussed in Section 3.7. Subsequently, isolation and active-device
Conclusion
Device isolation technology has substantially advanced during the past years. Besides ensuring undisturbed operation of millions of integrated devices, device isolation is now becoming increasingly an active part in device optimization, process integration and in specific applications. For example, using innovative concepts like EXTIGATE, ultra-dense DRAM and flash memories have been realized. We therefore conclude, that device isolation technology will further gain importance in the
Acknowledgements
The author would like to thank Dr. Preussger and Meinhard Obry for their encouragement, Dr. Kerber, Dr. Pompl, M. Füldner and C. Grünsfelder for valuable discussions as well as the entire staff of the ISTC-H84 for device fabrication. A portion of this work was performed within the ESPRIT 24115 ACE project and was supported by the European Union. Also, the excellent collaboration during the ACE Project with Dr. Bourenkov, Dr. Lorenz and Prof. Ryssel from the FhG IIS-B is acknowledged.
References (30)
- et al.
Thin Solid Films
(1992) - et al.
Philips Res Rep
(1970) - Fuse G, Ojawa H, Tateiwa K, Nakao I, Odanaka S, Fukumoto M, Iwasaki H, Ohzne T. IEDM Tech Dig...
- Davari B, Koburger C, Furukawa T, Taur Y, Noble W, Megdanis A, Warnock J, Mauer J. IEDM Tech Dig...
- Bryant A, Haensch W, Mii T. IEDM Tech Dig...
- et al.
IEEE ED
(1996) - Schwalke U, Kerber M, Koller K, Jacobs H. ESSDERC Conf Proc 1996. p....
- Schwalke U, Kerber M, Koller K, Ludwig B, Seidl A. Symp VLSI Tech...
- et al.
IEEE Trans Electron Dev
(1997) - Schwalke U, Füldner M, Zatsch W, Bothe K, Hadawi D, Janssen I, Schon P. Ext Abstr SSDM. Hiroshima, Japan. 1998. p....
Jpn J Appl Phys Part 1
IEEE EDL
IEEE EDL
Cited by (5)
Analysis of structural effect on mechanical stress at backside deep trench isolation using finite element method
2016, Microelectronic EngineeringCitation Excerpt :If the stress concentration is larger than fracture stress of silicon, a crack occurs from the BDTIs, and reliability gets damaged seriously.[8,9]. While previous works about the BDTI or DTI have focused on its performance as an image sensor like cross talk and leakage current, [4–7] such a mechanical reliability has never been addressed at the BDTI. We study the structural stability of the BDTI on the basis of a mechanical stress analysis using a TCAD simulation here.
Application of three-dimensional dislocation dynamics simulation to the STI semiconductor structure
2005, Materials Science and Engineering: ACitation Excerpt :Shallow trench isolation (STI) is a representative semiconductor structure and is used to realize electrical isolation between arrayed devices while maintaining high-density integration, unlike other structures such as LOCOS (local oxidation of silicon). However, it is well known that the thermal stress and the intrinsic stress of oxide films induce a high stress field and dislocation nucleation [3]. To meet such problems, transmission electron microscopy (TEM) observation is widely used as a powerful tool to infer the origin of dislocation nucleation, since this technique allows the shape of dislocation loop to be detected.
Near-ultraviolet Raman and micro-Raman analysis of electronic materials
2018, Applied Physics ReviewsApplication of dislocation dynamics simulation to the STI semiconductor structure
2004, Zairyo/Journal of the Society of Materials Science, Japan