Microdose Induced Data Loss on Floating Gate Memories

139

We discuss non-volatile memories (NVM) for space applications. The focus will be both on technologies and devices aimed at the mainstream commercial markets and on rad-hard devices. Commercial NVMs are very attractive for space designers due to their large size (tens of Gbits), even though they have several issues related to ionizing radiation. Rad-hard NVMs offer radiation hardness, but are available only in small size (few Mbits). Most of the emphasis in this review paper will be on the current dominant technology in the mainstream market: floating gate flash memories. A comprehensive discussion of total dose and single event effects results for a wide cross section of NVMs will be presented. Finally, we will conclude with a cursory glance at other emerging non-volatile technologies.Index Terms-Ferroelectric devices, flash memory, magnetoresistive devices, nonvolatile memory, phase change memory, radiation effects. I. OVERVIEWN ON-VOLATILE memories (NVMs) are used in a variety of space applications, as data or code memories, and are increasingly needed for future missions. This paper will discuss the technologies and devices available for non-volatile storage in the mainstream and in the rad-hard market, with a strong attention to the radiation sensitivity, which is of primary importance for harsh environments such as space.The work is organized as follows: we will first introduce the basic definitions and metrics concerning NVMs, then explore and compare the major storage concepts. In particular, charge-based, phase-change, ferroelectric, and magnetoresistive memories will be presented, with emphasis on density, performance, reliability, and radiation sensitivity. Architectural information concerning the array organization and peripheral circuitry will be given, highlighting known radiation issues and critical building blocks. A. Key Definitions and MetricsA non-volatile memory is a memory that retains information when powered off. Different terms are used to indicate the act of storing a digital value in a non-volatile memory bit, depending on the memory type and architecture. In charge-based storage, the terms program and erase are used to indicate the injection of Manuscript negative charge into and its removal from (or injection of positive charge into) the charge storage element, respectively. In the context of phase change memories, the terms set and reset are used to indicate the operations leading to the crystallization and amorphization of the phase-change material. For other memories, such as ferroelectric or magnetic, the term write is simply used, as for SRAMs and DRAMs. The association of these operations and terms with the stored digital value ("0" or "1") is arbitrary. In the following we will stick to the conventions most commonly used for each memory type.In addition to familiar parameters such as speed, power consumption, and density, two additional metrics are of main importance for non-volatile memories: retention and endurance. This is because wear-out is usually much stronger in non-volat...

Section: Radiation Effectsmentioning

confidence: 99%

Present and Future Non-Volatile Memories for Space

Gerardin

Paccagnella

2010

139

“…Yet, there are few reports on the evolution of radiation-induced FG cell errors as a function of the time elapsed after the exposure. Both in [13] and in [14], the Authors report that errors can anneal at room-temperature after heavy-ion exposure and they attribute the phenomenon to the removal of the radiation-induced positive charge in the oxides surrounding the Floating Gate. In [16] we presented an evolution of those ideas, assuming the occurrence of both charge loss and charge trapping (and its subsequent removal/neutralization).…”

Section: Introductionmentioning

confidence: 99%

Error Instability in Floating Gate Flash Memories Exposed to TID

Bagatin

Gerardin

Cellere

et al. 2009

We discuss new experimental results on the post-radiation annealing of Floating Gate errors in Flash memories with both NAND and NOR architecture. We investigate the dependence of annealing on the program level, linking the reduction in the number of Floating Gate errors to the evolution of the threshold voltage of each single cell. To understand the underlying physics we also discuss how temperature affects the number of Floating Gate errors

“…The mechanisms proposed to explain soft errors in the floating gate are connected with the phenomena of charge loss from the floating gate due to a transient conductive path across the tunnel oxide as well as charge trapping in the tunnel oxide [18,[23][24][25]. The heavy ion Fig.…”

Section: Fig 2(b)mentioning

confidence: 99%

Impact of Spacecraft Shielding on Direct Ionization Soft Error Rates for Sub-130 nm Technologies

Pellish

Xapsos

Stauffer

et al. 2010

Fig. 1: SiO2 sensitive volume (called SV1 in Table I) soft error rate relationships for several different space environments behind 2.5 mm of solid aluminum spherical shielding. All environments are derived from CRÈME96 with the exception of trapped protons and PSYCHIC. The deposited energy becomes the critical energy once a bin is chosen and the rate evaluated. The symbols are sparse to aid viewing -there are 240 bins per trace. The rate curve family for the silicon sensitive volume (called SV2 in Table I) is similar. Note that the PSYCHIC and trapped proton environments are cumulative fluences that were scaled by the inverse of the integration period to convert them to average rates.Abstract-We use ray tracing software to model various levels of spacecraft shielding complexity and energy deposition pulse height analysis to study how it affects the direct ionization soft error rate of microelectronic components in space. The analysis incorporates the galactic cosmic ray background, trapped proton, and solar heavy ion environments as well as the October 1989 and July 2000 solar particle events.Index Terms-soft error rate, direct ionization, radiation transport, space environment.