Wear-out, breakdown occurrence and failure detection in 18–25 Å ultrathin oxides

doi:10.1016/S0026-2714(01)00064-6

Microelectronics Reliability

Volume 41, Issue 7, July 2001, Pages 1035-1039

https://doi.org/10.1016/S0026-2714(01)00064-6 Get rights and content

Abstract

In this paper, a comprehensive description of the ultrathin oxide failure evolution is presented. For sub-25 Å, Hard BD is no longer hard. A complete description of the novel failure manifestation (progressive breakdown) is done. Associated wear-out is modelled and a physical mechanism is proposed. Finally, the relevance of the failure definition is discussed. It is a crucial point, to adopt a rigorous methodology for reliability prediction. It is concluded that, in the case of progressive breakdown, noise occurrence must be considered as the relevant time to failure.

Introduction

At the end of the 1980s Farmer [1] was the first to report the oxide failure evolution with the oxide thickness down scaling. He described that failure, in the sub-25 Å range, appears differently as compared to thicker oxides: the hard breakdown (HBD) becomes smoother and it is preceded by noise apparition.

Later, it was demonstrated that quasi-breakdown (QB) occurs for sub-5 nm gate oxides [2]. Its steady state current characteristic (several decades below the BD one) is shown to be independent of the oxide thickness [3], [4]. Its detection is based on current noise increase and low gate voltage current check for confirmation. Nevertheless with the continuous oxide thickness down scaling, direct tunnelling (DT) current dramatically increases, making failure detection and discrimination more difficult [5]. In this context, confusion between noisy BD and QB occurrence may induce physical misinterpretations and may mislead reliability predictions.

The aim of this paper is to clarify the wear-out and the failure modes for ultrathin oxides and to give some insights on the background mechanism explaining the evolution of the failure manifestation when scaling down the oxide thickness.

Section snippets

Experimental details

The devices used throughout this study are 18–26 Å ultrathin oxide N+/P MOS capacitors (area from 2 to $600,000 μm^{2}$ ) issued from a $0.12 μm$ CMOS technology. Constant voltage stress have been carried out under accumulation regime in the temperature range 25–125°C.

Novel failure manifestation

A new failure manifestation is typically illustrated in Fig. 1. It can be divided in three phases: (1) gate current noise suddenly appears, (2) gate current increases continuously and noisily, (3) either the current increases and reaches asymptotically its final value or a significant current jump occurs (case of the Fig. 1).

No clear significant (resp. slight) current jump which could be clearly associated to BD (resp. QB) can be recorded. In this paper, this novel phenomenon has been

Link with quasi breakdown

To investigate the localisation of the current flow for each event, light emission microscopy observations (Fig. 1, photos A, B and C) have been performed. It is clearly demonstrated first, that these different phases of the failure occurrence are related to the same defect path and second, that the observed current evolution is clearly associated to the wear-out of the same localised conduction path through the gate oxide. Notice that the final current steady state is the same as the HBD one

Wear-out modelling

Fig. 2 displays consecutive current characteristics measured on the same MOS device sample at different times from fresh up to broken down state. Observed current results in the fresh current (I_fresh) and the wear out current (I_w-o) superimposition. Important wear-out information is obtained since the BD path excess current conduction behaviour is clearly evidenced.

A striking feature is the similarity between fresh curve and all the others measured during wear-out (except BD one). This means

Relevant failure definition

The great question at this point is to determine what is the relevant breakdown time (T_bd) to consider in a gate oxide reliability assessment context. One point is that the noise appearance is the first manifestation of a physical sudden and localised wear-out, and therefore may be the most relevant failure definition.

In order to confirm this point, Fig. 5 displays the time to failure Weibull cumulative distributions obtained on a same set of device, considering three different failure

Wear-out mechanism

For a given time to failure, BD is found experimentally to become more progressive as the oxide thickness decreases, the gate voltage decreases and the temperature increases (Fig. 7). In addition, progressiveness is not impacted by device area. This is understandable since PBD is a localised phenomenon.

Stored energy (Eq. (2)) in the capacitor is supposed to influence failure mode determination [11], [12]. $U= 12 A_{w-o} ε_{ox} E_{ox}^{2} T_{ox}$ E_ox being the oxide field, T_ox the oxide thickness, ε_ox the oxide

Conclusion

For sub-25 Å thick oxides, a novel failure manifestation appears. Each part of this wear-out (noise occurrence, current increase and current jump) is found to be part of the same localised phenomenon. Thus, it only can be attributed to a new BD manifestation and cannot be ascribed to a QB one since BD and QB are found to be different phenomena. This progressive BD is modelled as a continuous oxide thinning on a localised conduction path area.

This work has clearly evidenced that rigorous failure

Acknowledgements

The authors would like to acknowledge S. Bruyère, D. Roy (ST Microelectronics, Central R&D) and W.J. Toren (Phillips/Crolles) for helpful discussions. This work is supported by ULTIMOX RMNT project.

References (12)

Farmer KR, Saletti R, Buhrman RA. Appl Phys Lett...
Okada K, Kawasaki S, Hirofuji Y. SSDM, 1994. p....
Bruyère S, Vincent E, Ghibaudo G. IRPS, 2000. p....
Miranda E, Suñe J, Rodriguez R, Nafria M, Aymerich X. IEEE Elec Dev Lett...
Miranda E, Suñe J, Rodriguez R, Nafria M, Aymerich X. SSE 43, 1999. p....
Suehle JS, Vogel EM, Wang B, Bernstein JB. IRPS, 2000. p....

There are more references available in the full text version of this article.

Cited by (46)

Ultra-thin dielectric breakdown in devices and circuits: A brief review
2015, Microelectronics Reliability
Citation Excerpt :
In addition, since the gate leakage due to BD (IG_BD) in ultra-thin oxide is few orders of magnitude smaller than that in thicker oxide, the first BD event in thin oxides might not cause failure of device. As a result, significant research efforts in recent years have been devoted to understanding post-BD phenomenon, such as SBD, progressive BD (PBD), and HBD [7–9]. On the other hand, the emergence of advanced devices, such as high-k metal gate (HKMG) transistors and magnetic tunnel junctions (MTJ) [10–15], has added new dimensions to TDDB study [15–25].
Time-dependent dielectric breakdown (TDDB), in which the traps in oxide bulk form a conducting path under application of stress voltage for long period of time, has emerged as one of the important sources of performance degradation in advanced devices. In this paper, we give an overview of the recent progress in the understanding of ultra-thin dielectric breakdown in devices and consider its impact at the circuit-level. From the device point of view, the breakdown (BD) phenomenon, including the BD statistics, trap generation models, and BD evolution in ultra-thin dielectric are presented followed by the recent studies on TDDB in high-k metal gate (HKMG) devices and magnetic tunnel junction (MTJ) memories. On the circuit side, we explore methodologies for circuit lifetime assessment, the impact of TDDB on circuit performance degradation, and design techniques to improve circuit reliability.
Burn-in considering yield loss and reliability gain for integrated circuits
2011, European Journal of Operational Research
This paper presents burn-in effects on yield loss and reliability gain for a lifetime distribution developed from a negative binomial defect density distribution and a given defect size distribution, after assuming that the rate of defect growth is proportional to the power of the present defect size. While burn-in always results in yield loss, it creates reliability gain only if either defects grow fast or the field operation time is long. Otherwise, burn-in for a short time could result in reliability loss. The optimal burn-in time for maximizing reliability is finite if defects grow linearly in time and is infinite if defects grow nonlinearly in time. The optimal burn-in time for minimizing cost expressed in terms of both yield and reliability increases in the field operation time initially but becomes constant as the field operation time is long enough. It is numerically shown that increasing mean defect density or defect clustering increases the optimal burn-in time.
A novel approach to characterization of progressive breakdown in high-k/metal gate stacks
2008, Microelectronics Reliability
Degradation of high-K LA<inf>2</inf>O<inf>3</inf> gate dielectrics using progressive electrical stress
2005, Microelectronics Reliability
The evolution of the leakage current in high-K lanthanum oxide films in MOS devices caused by the application of progressive electrical stress is investigated. The degradation method consists in performing successive voltage sweeps using an ever increasing voltage range with the aim of generating incremental damage to the structure in a controlled manner. We show that the total current flowing through the device can be thought of as formed by two parallel components, one associated with the tunneling mechanism and the other one associated with diode-like conduction. This latter component evolves with applied stress. It is shown the importance of considering series and parallel resistances in order to account for the right shape of the conduction characteristics. Analytical expressions for both current contributions suitable for all stages of degradation and bias conditions are provided.
Impact of Fowler-Nordheim and channel hot carrier stresses on MOSFETs with 2.2nm gate oxide
2005, Microelectronic Engineering
We studied the degradation of NMOSFETs with 2.2nm gate oxide subjected to Fowler-Nordheim and channel hot carrier stresses. We highlighted the differences between the two types of stresses, in particular concerning the behavior of the gate current. Before hard breakdown, Fowler-Nordheim stressed devices showed pre-breakdown and/or soft-breakdown, while progressive breakdown was much more common with channel hot carrier stresses, especially at V_gs = V_ds/2. The impact of the stresses on the transfer and output characteristics of the devices was then considered. Finally, the effect of the degradation on adjacent, nominally identical MOSFETs was studied, showing potential issues for circuits that rely on matching.
Statistics of progressive breakdown in ultra-thin oxides
2004, Microelectronic Engineering
Electrical breakdown manifests itself in ultra-thin oxides by a gradual and systematic rise in gate leakage current. On average, this progressive breakdown is shown to increase the gate leakage exponentially with time, with the time constant exponentially dependent on the gate voltage. Because thin oxides do not instantaneously short at the commencement of breakdown, the definition of oxide failure may be altered from the first jump in gate current to the leakage value that disrupts circuit operation. Depending on the ratio of the defect production rate to the defect growth rate, multiple breakdowns may form in a single device. A Monte Carlo simulator combines these results and calculates the oxide failure distribution with failure defined by a leakage level. The simulations reveal that updating the definition of oxide failure increases the projected reliability by many orders of magnitude.

View all citing articles on Scopus

View full text