Median Time To Failure Research Articles

Electromigration in Al based stripes: Low frequency noise measurements and MTF tests

Traditional [Median time to Failure (MTF)] and non traditional (Noise measurements) techniques have been used to characterize four types of Al based samples. Relatively weak stress conditions, which cause negligible modifications to the sample structure, have been used for noise measurements, whereas more accelerated stress conditions have been used for lifetime tests. Quite different results have been obtained by the two types of characterization. In fact, the experimental data showed that the noise measurements are useful to evaluate the activation energy at relatively weak stress conditions, but cannot be used to foresee the behavior of the samples at strongly accelerated stress conditions. This is particularly true in the case of samples containing copper which probably undergo significant microstructural modifications as the temperature goes up. Moreover, the results of a simple computer simulation which show the dramatic effect on the Time to Failure of the spreading of the values of the thermal resistance of the samples are reported.

Prediction and Measurement of Temperature Fields in Silicon-on-Insulator Electronic Circuits

Field-effect transistors (FETs) in conventional electronic circuits are in contact with the high-thermal-conductivity substrate. In contrast, FETs in novel silicon-on-insulator (SOI) circuits are separated from the substrate by a thermally resistive silicon-dioxide layer. The layer improves the electrical performance of SOI circuits. But it impedes conduction cooling of transistors and interconnects, degrading circuit reliability. This work develops a technique for measuring the channel temperature of SOI FETs. Data agree well with the predictions of an analytical thermal model. The channel and interconnect temperatures depend strongly on the device and silicon-dioxide layer thicknesses and the channel–interconnect separation. This research facilitates the thermal design of SOI FETs to improve circuit figures of merit, e.g., the median time to failure (MTF) of FET–interconnect contacts.

High-Reliability Copper Interconnects through Dry Etching Process

A modified high-temperature dry etching technique, which enables anisotropic patterning with a high etching selectivity and self-aligned passivation of a sidewall of an interconnect simultaneously, has been developed for fabrication of sub-quarter-micron Cu interconnects. Resistivities of the resulting Cu interconnects are in the range of 1.7 to 2.2 µΩ· cm for the linewidth of 0.2-3.0 µ m. As a result of electromigration (EM) tests, it has been observed that median time to failure (MTF) of the Cu interconnects depends on their linewidth. This behavior is considered to be caused by their grain structure, such as a bamboo-type structure for linewidths narrower than 0.3 µ m. In comparison with a MTF of a conventional Al-1%Si line, these Cu interconnects have at least 100 times longer lifetime. Activation energy for EM damage of the 0.7-µ m-wide line is 0.88 eV.

The Influence of Grain Structure on the Reliability of Narrow Al-Based Interconnects

AbstractThe work reported here concerns the effect of grain structure on electromigration failure in pure Al and Al-2wt.%Cu-lwt.%Si lines. The grain structures of fine lines were controlled by annealing after patterning to promote the formation of "bamboo" structures. Significant improvements in the median time to failure (MTF) and the deviation of the time to failure (DTF) were observed with the development of near-bamboo structures with polygranular-segment lengths shorter than ~ 5 μm. The most common failure sites are voids or slits across bamboo grains at the upstream ends of polygranular segments. The time-to-failure decreases with the polygranular segment length, and can be significantly enhanced by controlling the grain structure.

Effectiveness and Reliability of Metal Diffusion Barriers for Copper Interconnects

AbstractCu has two advantages over Al for sub-quarter micron interconnect application: (1) higher conductivity and (2) improved electromigration reliability. However, Cu diffuses quickly in SiO2and Si, and must be encapsulated. Polycrystalline films of Physical Vapor Deposition (PVD) Ta, W, Mo, TiN, and Metal-Organo Chemical Vapor Deposition (MOCVD) TiN and Ti-Si-N have been evaluated as Cu diffusion barriers using electrically biased-thermal-stressing tests. Barrier effectiveness of these thin films were correlated with their physical properties from Atomic Force Microscopy (AFM), Transmission Electron Microscopy (TEM), Secondary Electron Microscopy (SEM), and Auger Electron Spectroscopy (AES) analysis. The barrier failure is dominated by “micro-defects” in the barrier film that serve as easy pathways for Cu diffusion. An ideal barrier system should be free of such micro-defects (e.g., amorphous Ti-Si-N and annealed Ta). The median-time-to-failure (MTTF) of a Ta barrier (30 nm) has been measured at different bias electrical fields and stressing temperatures, and the extrapolated MTTF of such a barrier is > 100 year at an operating condition of 200C and 0.1 MV/cm.

Planarized Copper Interconnects by Selective Electroless Plating

Electroless deposition of planarized copper in SiO 2 trenches has been carried out using Pd/Si plasma immersion ion implantation (PIII) or Pd 2 Si deposition to form a seed layer at the bottom of trenches. Electrical resistivity of the plated Cu is ≤ 2 μΩ-c-m for both types of seeding layers. For the PIII seeding method, we found a threshold Pd dose of 2 ×10 14 /cm 2 is required to initiate Cu plating, and RBS analysis confirms intermixing of Pd, Si, and SiO 2 improves the adhesion of the plated Cu to SiO 2 . Electromigration tests show both void and hillock formation under accelerated current stress testing, with an activation energy of 0.8 eV for interconnect open failure. The DC and pulse-DC median-time-to-failure (MTF) of plated Cu are found to be about two orders of magnitude longer than that of AI-2%Si at 275°C, and about four orders of magnitude longer than that of AI-2%Si when extrapolated to room temperature. Pulsed DC electromigration stressing exhibits a transition from low to high frequency behavior around 900 kHz, indicating a vacancy relaxation time much shorter than that of Al. For bipolar AC stressing, the ratio MTF Ac / MTF DC of Cu is much smaller than that of A1–2%Si, indicating a different void recovery mechanism for Cu.

Electromigration in Bicrystal Al Lines

We have developed a new experimental technique to study electromigration in bicrytal Al lines as a function of the type and location of the grain boundary as well as the testing temperature. The failure times of these lines are found to be lognormally distributed. The median time to failure (MTTF) depends more strongly on the boundary orientation than the type of grain boundary. The dependence of lifetimes on the type and orientation of grain boundaries, the location and appearance of the failure sites, and the measured activation energy (E.) of 0.94eV suggest that both interfacial and grain boundary diffusion contribute to failure in bicrystal lines, and likely in bamboo and near-bamboo lines as well.

Planarized Copper Interconnects by Selective Electroless Plating

ABSTRACTElectroless deposition of planarized copper in SiO2 trenches has been carried out using Pd/Si plasma immersion ion implantation (PIII) or Pd2Si deposition to form a seed layer at the bottom of trenches. Electrical resistivity of the plated Cu is < 2 μΩ-cm for both types of seeding layers. For the PHI seeding method, we found a threshold Pd dose of 2 ×l014/cm2 is required to initiate Cu plating, and RBS analysis confirms intermixing of Pd, Si, and SiO2 improves the adhesion of the plated Cu to SiO2. Electromigration tests show both void and hillock formation under accelerated current stress testing, with an activation energy of 0.8 eV for interconnect open failure. The DC and pulse-DC median-time-to-failure (MTF) of plated Cu are found to be about two orders of magnitude longer than that of Al-2%Si at 275°C, and about four orders of magnitude longer than that of Al-2%Si when extrapolated to room temperature. Pulsed DC electromigration stressing exhibits a transition from low to high frequency behavior around 900 kHz, indicating a vacancy relaxation time much shorter than that of Al. For bipolar AC stressing, the ratio MTFAC / MTFDC of Cu is much smaller than that of Al-2%Si, indicating a different void recovery mechanism for Cu.

Atomistic and computer modeling of metallization failure of integrated circuits by electromigration

Nonstationary and stationary concentration profiles of vacancies in a grain boundary are calculated for the case of electromigration and various conditions of vacancy formation and annihilation. If supersaturation of vacancies is assumed for void formation, current densities necessary for voiding have to be larger than a critical value jcrit, which is related to the length l of the grain boundary by the relation ljcrit=const. An estimation of the value of the constant in the last equation is in agreement with experimental values. An approximate equation for the median time to failure (MTF) is derived from the model for grain boundaries conducting the flux of matter parallel or in series. The results explain many features of experimental studies. First values of MTF for a network of grain boundaries obtained by computer simulations are presented as well, and the effect of resistance heating and nonstationarity on the current exponent is discussed in some detail.

The effect of microstructure on the resistance to electromigration of AlCu thin film conductors

Aluminum-copper alloy films were prepared by vacuum evaporation under varying conditions of deposition rate and substrate temperature. They were subsequently subjected to electromigration testing using the median time to failure (MTF) and temperature-ramp resistance analysis to characterize electromigration (TRACE) methods and the resulting activation energies for electromigration were extracted. It was found that certain microstructural configurations affected the beneficial effects of copper on the resistance of thin films to electromigration. The magnitude of the activation energies ranged from 0.25 eV up to 0.95 eV. The differences in the activation energies Q were related to the deposition conditions, the films' grain size distributions, the copper content of the films, the Al 2Cu precipitate distributions in the films, and the possible presence of Guinier-Preston zones. The Avrami analysis of the resistometric data for the lifetime experiments was also employed. It was found that a single failure mechanism dominated over the temperature range of the MTF experiments, but this mechanism was dependent on the deposition conditions.

An extension of probabilistic simulation for reliability analysis of CMOS VLSI circuits

The probabilistic simulation approach is extended to include the computation of the variance waveform of the power/ground current, in addition to its expected waveform. The focus is on the problem of estimating the median time-to-failure (MTF) due to electromigration (EM) in the power and ground buses of CMOS circuits. Theoretical results that quantify the relationship between the MTF and the statistics of the stochastic current are presented. This leads to a more accurate estimate of the MTF that requires both the expected and variance waveforms. A novel technique is then presented to compute the variance waveform for CMOS circuits, which has been incorporated into the probabilistic simulator CREST. Results of this implementation demonstrating efficiency and accuracy on a number of circuits are provided. The authors use these results to study the importance of the variance waveform by estimating its contribution to the MTF relative to that of the expected waveform.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

1/f/sup alpha / noise and fabrication variations of TiW/Al VLSI interconnections

Noise measurements were utilized to detect differences in fabrication processes on the layered TiW/Al metallization system. A dual-channel AC bridge system was used to measure the current noise of packaged 1.8- mu m-wide thin-film interconnections fabricated by a variety of processing parameters. The quick, nondestructive noise measurements were able to discriminate relative film group reliabilities and correlated well with results obtained from more conventional median-time-to-failure (MTF) testing.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Grain size dependence of electromigration-induced failures in narrow interconnects

Measurements of the median time to failure (MTTF) and deviation in the time to electromigration-induced failure (DTTF) of Al alloy thin-film lines are reported. As the ratio of the linewidth to the grain size decreases, MTTF decreases to a minimum and then increases exponentially. DTTF continuously increases. We show that serial and parallel failure unit models can be used to explain the grain size and linewidth dependence of the MTTF and DTTF for interconnects. We further note that extrapolation to low cumulative failures based on serial failure models must be based on knowledge of the failure statistics of individual units.

The deposition and film properties of reactively sputtered titanium nitride

Titanum nitride films were prepared by reactive d.c. magnetron sputtering from a titanium target in a working gas mixture of Ar + N 2. Film preparation was examined with respect to kinetics during reactive sputtering. Rutherford backscattering, Auger electron spectroscopy, transmission electron microscopy and electron diffraction were used to determine film composition, impurity levels and microstructure. The stoichiometric TiN films were found to possess a single-phase f.c.c. structure and demonstrated good diffusion barrier integrity in an Al/TiN/Si structure at temperatures below 550 °C. Interposition of a thin 60 Å titanium layer at the TiNSi interface yielded low contact resistance to p +- and n +-Si regions. The limitation of reactively sputtered TiN diffusion barrier layers was determined to be their effect on the electromigration of the overlying aluminum, an effect which has not previously been reported. The insertion of a TiN layer between the aluminum and the underlying SiO 2 shifted the electromigration failure mode from open- to short-circuit formation and reduced the median time to failure (MTF) by more than an order of magnitude. The MTF of the composite Al/TiN conductor was a factor of 5 less than that for a comparable Al/TiW conductor. Preliminary experimentation with layered Al/Ti conductors revealed the electromigration failure mode also to be short-circuit formation.

A comparison of the median time to failure with temperature-ramp resistance analysis to characterize electromigration to determine electromigration kinetic parameters

Temperature-ramp resistance analysis to characterize electromigration (TRACE) can yield electromigration damage (EMD) kinetic parameters in times as short as a day. Until now, there has been no systematic comparison of kinetic parameters determined by TRACE and median time to failure (MTF). In this work, two groups of Al-4%Cu conductors were tested by both techniques. The average activation energy for EMD was determined by the TRACE technique to be 0.69 eV. This value is in excellent agreement with the value obtained from life testing. The t 50 calculated from TRACE is shorter than the measured MTF value by a factor of approximately ten.

SPIDER -- A CAD System for Modeling VLSI Metallization Patterns

A system of CAD programs, called SPIDER, for ensuring adequate current-carrying capacity in VLSI circuits has been developed. The approach is hierarchical, and it automates and simplifies many of the tasks previously performed by the circuit designer. The system converts transient current waveforms into dc electromigration equivalent values, and includes an algorithm for determining the line width adjustments necessary for meeting specified median-time-to-failure (MTF) requirements. SPIDER also computes voltage drops with respect to a reference node and calls upon an optimization algorithm for correcting violations of a voltage drop specification.

In‐Source Al‐0.5%Cu Metallization for CMOS Devices

This study examines the use of radiation‐free RF induction‐source evaporated (In‐Source®) Al‐0.5%Cu as a viable metallization scheme for CMOS devices. The fact that it is radiation free offers ways of optimizing the thermal cycle so as to achieve desirable device and electromigration characteristics. Wafers fabricated with 3.5 μm design rules have shown no threshold shifts or changes in the transistor gain at sintering temperatures up to 450°C, in either or , for both N‐ and P‐channel devices, confirming the absence of deposition‐induced fixed interface charges. Wafers processed with 2.5 μm design rules have exhibited low and well‐behaved values of contact resistance to N+ and P+ diffusions, at sintering temperatures in the range 300°–400°C in either or , with no evidence of junction penetration. Electromigration measurements, gathered over a wide range of processing conditions, have yielded values of median time to failure (MTF) that were observed to improve with narrower linewidths, higher sintering temperatures, and in the presence of a plasma passivation coating. In typical device operation conditions of 80°C and , the extrapolated MTF value of , for 3.0 μm wide In‐Source Al‐0.5%Cu lines is found to exceed the value of reported for E‐Gun Al‐0.5%Cu.

The reliability of (AlGa)As CW laser diodes

This paper reviews the major factors bearing on the reliability of (AlGa)As CW laser diodes. The degradation modes of facet mirror damage, contact degradation, and internal damage are discussed in terms of our present knowledge of their effects on device performance, their origin, and their reduction or elimination. Detailed results are presented for oxide-defined stripe-contact lasers and, although reliability results are a strong function of fabrication technology, there are many issues common to all fabrication technologies, and these are emphasized. Lasers have been operated in our laboratory for more than 40 000 h with extrapolations indicating a median time to failure (MTTF) between 105and 106h. In these lasers both facet damage and contact degradation appear to be under control and internal damage remains the dominant failure mechanism. While still not well documented, for internal damage a so called activation energy of 0.7 eV may be useful for high temperature accelerated lifetests. Most of the reliability data deals with threshold current increase, however, shifts in far-field pattern and changes in laser modulation characteristics, such as self-sustained oscillations, may affect laser performance in real systems.

Median-time-to-failure (MTF) of an L-Band power transistor under RF conditions : W. E. Poole and L. G. Walshak. Proc. IEEE Reliab. Physics Symp. 109 (1974)

Median-time-to-failure (MTF) of an L-Band power transistor under RF conditions : W. E. Poole and L. G. Walshak. Proc. IEEE Reliab. Physics Symp. 109 (1974)

Electromigration of Ti–Au thin-film conductors at 180° C

A total of 394 test stripes on sapphire substrates were subjected to constant current density [(2−8)×106A/cm2] at 180 °C. Resistance was measured periodically during the test. Sixty-eight percent of the samples were tested to open-circuit failure. For the remainder, lifetimes were predicted on the basis of the time-on-test and the lifetime statistics of the actual failures. The principal results are as follows: (i) Lifetime varied about as the inverse fourth power of the current density, and was almost independent of the stripe cross-sectional area per se. (ii) The initial slope of the resistance-time curve is highly correlated with, and varies inversely as, the lifetime. (iii) The distribution of failure times in this batch of test stripes was very broad, some 10% failing at times less than one-tenth the median time to failure (MTF), and about 4% at times greater than 10× the MTF. (iv) Estimates of performance under various conditions of service have been made. Operation at current densities above the mid 105 A/cm2 range at 100 °C are not recommended for high-reliability applications.