Rolf-Peter Vollertsen

Experimental evidence of T/sub BD/ power-law for voltage dependence of oxide breakdown in ultrathin gate oxides

In this paper, we present experimental evidence on the voltage-dependence of the voltage acceleration factors observed on ultrathin oxides from 5 nm down to /spl sim/1 nm over a wide range of voltages from /spl sim/2 V to 6 V. Two independent experimental approaches, area scaling method and long-term stress, are used to investigate this phenomenon. We show the exponential law with a constant voltage-acceleration factor violates the widely accepted fundamental breakdown property of Poisson random statistics while the voltage-dependent voltage acceleration described by an empirical power-law relation preserves this well-known property. The apparent thickness-dependence of voltage acceleration factors measured in different voltage ranges can be nicely understood and unified with these independent experimental results in the scenario of a voltage-driven breakdown. In the framework of the critical defect density and defect generation rate for charge-to-breakdown, we explore the possible explanation of increasing voltage acceleration factors at reduced voltage by assuming a geometric model for the critical defect density.

Understanding soft and hard breakdown statistics, prevalence ratios and energy dissipation during breakdown runaway

Since some MOS digital circuits could remain functional after gate oxide breakdown (BD) provided that the post-BD resistance is high enough (1), the separate consideration of soft (SBD) and hard (HBD) breakdown events is necessary to set up an adequate application-specific reliability assessment methodology. In this work we deal with two relevant issues related to this problem. First, we study the statistics of SBD and HBD and their relation to the first-event BD statistical distribution as a function of their prevalence ratios. Second, we consider the modeling of the BD runaway as a means to determine these prevalence ratios as a function of stress conditions and device geometry.

Weibull breakdown characteristics and oxide thickness uniformity

In this work, we investigated both experimentally and numerically the impact of macroscopic oxide thickness uniformity on Weibull breakdown characteristics for both Weibull parameters, namely, the characteristic times and Weibull slopes over a wide range of oxide thicknesses. We report the abnormal characteristics of the Weibull time-to-breakdown distributions and non-Poisson area scaling behavior observed on ultrathin oxides. Two numerical methods using the parameters obtained from two independent sets of experimental results are developed to quantitatively explain these effects in the context of current modulation due to oxide thickness variation. The relationship between time-to-breakdown and charge-to-breakdown distributions has been clarified and established. It is found that without proper treatment of these effects, the use of Weibull slopes at higher failure percentiles can lead to erroneous and pessimistic reliability projection. Furthermore, me perform a detailed full-scale Monte Carlo analysis to evaluate the impact of thickness variation on low-percentile breakdown distributions and their sensitivity to the thickness dependence of the times-to-breakdown and Weibull slopes.

New global insight in ultrathin oxide reliability using accurate experimental methodology and comprehensive database

In this paper, we critically examine several important experimental aspects concerning ultrathin oxide reliability. The statistical nature of breakdown measurements and the impact on data interpretation is discussed. Thickness dependence of Weibull slopes and its impact on reliability projection is reviewed. We also investigate the voltage-dependent voltage acceleration using two independent experimental methods over a wide range of oxide thickness values. Within the framework of a general defect generation model, we explore the possibility of a voltage-dependent defect generation rate to account for the increase in voltage acceleration with decreasing voltages. Using direct experimental results, we clarify that strong temperature dependence found on ultrathin oxides is a voltage effect, not a thickness effect as previously suggested, In the context of voltage-dependent voltage acceleration, we experimentally resolve various seemingly contradicting and confusing observations such as temperature-independent voltage acceleration and non-Arrhenius temperature dependence found on ultrathin oxides. Finally, we provide a global picture for time-to-breakdown in voltage and temperature domain constructed from two important empirical principles based on comprehensive experimental database.

An introduction to fast wafer level reliability monitoring for integrated circuit mass production

The continuous verification of process reliability is essential to semiconductor manufacturing. The tool that accomplishes this task in the required short time is the fast wafer level reliability monitoring (fWLR). The basic approaches for this task are described in this introductory overview. It summarizes sampling plans, discusses the feasibility of using fWLR for screening and describes the data assessment and application of control cards. Beyond these general topics many of the fWLR stress methods are described in detail: Dielectric stressing by means of an exponential current ramp is compared to ramped voltage stress. Especially for thin oxides the methods differ regarding the soft breakdown detection and the time they consume. Another task of fWLR is the detection of plasma induced damage, which can be achieved by applying a revealing stress to MOSFETs with antenna. The design challenges of the structures and the test method as well as the data assessment are described in detail. An important section deals with fWLR for interconnects. In this section the appropriate test structures (including thermal simulations) are illustrated and fast electromigration stresses are discussed and the details of standard wafer level electromigration accelerated test (SWEAT) are included. For contacts and vias a simple method to check reliability is presented. Finally the monitoring of device reliability is treated. It is shown that using indirect parameters that correlate well to standard parameters such as the drain current can be beneficial for fWLR. For both, the interconnects and the devices, it is essential to have locally heated test structures in order to keep the stress time low. The detection and verification of mobile ions can also be performed with such a self-heated structure. For the described methods examples are given to illustrate the usefulness.

A novel multi-point NBTI characterization methodology using Smart Intermediate Stress (SIS)

In recent literature several measurement methods were introduced to characterize the Vth-degradation due to NBTI considering the recovery phenomenon. To our knowledge each method has a severe problem or at least a significant disadvantage. Either there are long delay times, the accuracy is not satisfactory or it is not possible to implement the method with customary equipment. A compromise is to perform a one point measurement in the subthreshold region and calculate Vth based on the assumption that the subthreshold slope is not affected by NBTI. In this paper we disprove the universality of this assumption. Vth determination using a one point measurement can lead to imprecise values. This extraction method disregards changes of the subthreshold slope due to NBTI, however a change of the slope impacts the extracted Vth. We clearly demonstrate this effect with our measurements. We introduce a new smart Vth extraction methodology offering both shortest possible delay times with customary equipment and consideration of NBTI-impact on subthreshold slope.

A Study of NBTI and Short-Term Threshold Hysteresis of Thin Nitrided and Thick Non-Nitrided Oxides

Negative bias temperature instability (NBTI) degradation and recovery have been investigated for 7-50-nm non-nitrided oxides and compared to thin 1.8- and 2.2-nm nitrided oxides from a dual work function technology. A wide regime of stress fields from 2.5 to 10 MV/cm has been covered. Thermal activation has been studied for temperatures from 25 degC to 200 degC. The NBTI effect for the nitrided oxide is larger than for non-nitrided oxides. The percentage of threshold shift V th which is ldquolostrdquo during a long measurement delay-which is the quantity leading to curved V th versus stress-time curves and to errors in extrapolated lifetimes-is about equal for nitrided or thick non-nitrided oxides. The fraction of recovered V th is strongly dependent on stress time but only weakly dependent on stress field. Recovery in thick oxides leads to exactly the same problems as for non-nitrided oxides, and clearly, a fast measurement method is needed. The effect of short-term threshold shifts has been studied for extremely short stress times down to 200 ns.

Comprehensive physics-based breakdown model for reliability assessment of oxides with thickness ranging from 1 nm up to 12 nm

The state-of-art understanding on the TBD voltage acceleration models in direct tunneling (DT) and Fowler- Nordheim (FN) regimes is thoroughly and carefully reviewed including recent work on thin oxides as well as historical publication database for thick oxides. The field-driven TBD exponential law is found to be inconsistent with many experimental findings. We present a comprehensive physicsbased breakdown model, which separately takes the roles of tunneling current and defect generation efficiency into account, and it is consistent with many experimental findings for thickness from 1.0nm to 12nm. With these new advanced understandings, we can now resolve many controversies surrounding TBD voltage acceleration models for SiO2-based dielectrics. Finally, a practical solution of acceleration model for TDDB qualification is proposed.

A reliable and accurate approach to assess NBTI behavior of state-of-the-art pMOSFETs with fast-WLR

NBTI is a key challenge of todays technologies and could be assessed so far only by relative long stress durations. The On-the-fly characterization seems to be a proper method for fast-WLR, but shows also some problems. This work describes for the first time the application of OTF in combination with self-heating test structures and a method to correct the initial value without special equipment. Challenges and different solutions are introduced and discussed. We compare data from very fast non standard measurements with fWLR data acquired with regular test equipment. The insight we gained suggests that fWLR provides a suitable means for fast NBTI monitoring.

Voltage acceleration and t63.2 of 1.6–10 nm gate oxides

Abstract Gate oxide reliability data collected over a considerable period of time were compiled to assess the voltage acceleration and the time to breakdown as function of oxide thickness. These data cover a range from 1.6 to 10 nm and can be used as benchmark for technologies that are still using gate oxide in this thickness range. The data form well-defined bands for each of the voltage acceleration models. The functional dependence of the parameter on oxide thickness depends strongly on the voltage acceleration model. The accuracy of the voltage acceleration parameters determined for the different acceleration models is studied. The time to breakdown at one voltage spans many time-decades if the data covering the entire thickness range are plotted in one graph. Therefore, the use of a model-free value, the voltage to get 63.2% breakdown at a certain fixed time, is proposed for plotting the data taken in the wide oxide thicknesses range, instead of normalizing the time to breakdown to a certain voltage using one of the voltage acceleration models. Based on the results a self-consistent test of the voltage acceleration models is introduced. This parameter also supports the tbd power law and therefore the hydrogen release model when plotting the voltage acceleration parameter of the exp(V)-model versus the inverse model-free gate voltage to get 63.2% breakdown at a fixed time.