Steffen Kaldor

Comparison of Cu electromigration lifetime in Cu interconnects coated with various caps

Electromigration in Cu Damascene lines with bamboo-like grain structures, either capped with Ta/TaN, SiNx, SiCxNyHz layers, or without any cap, was investigated. A thin Ta/TaN cap on top of the Cu line surface significantly improves electromigration lifetime when compared with lines without a cap and with lines capped with SiNx or SiCxNyHz. The activation energy for electromigration increased from 0.87 eV for lines without a cap to 1.0–1.1 eV for samples with SiNx or SiCxNyHz caps and to 1.4 eV for Ta/TaN capped samples.

In situ study of void growth kinetics in electroplated Cu lines

An in situ electromigration apparatus was used to study the kinetics of void growth in unpassivated, electropolated copper damascene lines. Voids were observed to grow by consuming grains in a stepwise fashion, either by grain thinning or by an edge displacement mechanism. Surface diffusion was found to be the primary diffusion path for void growth. In addition, grain boundaries provided a secondary path for copper diffusion in polycrystalline structures and nucleation sites for void growth in bamboo structures. Void growth rate was measured as a function of sample temperature and linewidth using a scanning electron microscope. An electromigration activation energy of 0.9±0.1 eV was determined for the copper voiding process. The effect of linewidth on void growth rate was also investigated and found to be negligible, consistent with a surface-diffusion dominated model for void growth. The in situ apparatus also made it possible to directly correlate changes in electrical resistance with physical changes t...

Effects of overlayers on electromigration reliability improvement for Cu/low K interconnects

Electromigration in Cu Damascene lines capped with either a CoWP, Ta/TaN, SiN/sub x/, or SiC/sub x/N/sub y/H/sub z/ layer was reviewed. A thin CoWP or Ta/TaN cap on top of the Cu line surface significantly reduced interface diffusion and improved the electromigration lifetime when compared with lines capped with SiN/sub x/ or SiC/sub x/N/sub y/H/sub z/. Activation energies for electromigration were found to be 2.0 eV, 1.4 eV, and 0.85-1.1 eV for the Cu lines capped with CoWP, Ta/TaN, and SiN/sub x/ or SiC/sub x/N/sub y/H/sub z/, respectively.

Characterization of substrate/thin-film interfaces with x-ray microdiffraction

The (buried) interface between a polycrystalline Al thin-film feature and its substrate (single crystal Si) was characterized with x-ray microdiffraction. Using a focused x-ray beam (effective spot size on the specimen ∼2×12 μm) with the Si 004 reflection, topographic images of the Si around and under the metallization feature were constructed. Comparison with shear-lag model calculations indicate that the interface is not fully coupled despite the absence of surface cracks.

Deformation field in single-crystal fields semiconductor substrates caused by metallization features

The results of an x-ray microdiffraction study of the deformation field surrounding Ni thin film pads on a 111-type Si wafer are reported. The strain fields were mapped by measuring the Si 333 reflection intensities over an area containing several pads. The positions of the pads were simultaneously determined by recording the Ni Kα fluorescence as a function of position. The results indicate that, contrary to the results from analytical solutions and finite-element models, the position of maximum strain contrast is slightly outside the pad edge.

Observation of local tilted regions in strain-relaxed SiGe/Si buffer layers using x-ray microdiffraction

The microstructure of strain-relaxed Si1−xGex/Si films that relaxed by different dislocation nucleation mechanisms has been investigated using x-ray microdiffraction with a diffracted beam footprint of 1 μm×5μm. Intensity variations in the x-ray microtopographs of samples having step-graded intermediate layers, which relaxed by dislocation multiplication, are due to the presence of local tilted regions which are larger in area than the diffracted x-ray beam. In contrast, microtopographs of uniform composition layers, which relaxed by surface roughening and subsequent random dislocation nucleation, show little intensity contrast as the local tilted regions in these samples are much smaller than the diffracted x-ray beam. The difference in microstructure arises from the different distributions of 60 ° misfit dislocations in these two types of samples.

Real-time x-ray microbeam characterization of electromigration effects in Al(Cu) wires

We report real-time, in situ x-ray microbeam measurements of electromigration-induced Cu redistribution, and the concurrent local stress variation in Al(Cu) wires. The data, which were obtained by combining x-ray microtopography with energy-dispersive fluorescence analysis, encompass both the early and late stages of electromigration as well as the postrelaxation stage at high temperature with the current turned off. We observe that both Cu concentration and stress values show unexpected local variations that may reflect the effect of local configuration such as film–substrate interface integrity or microstructure.

Divergence effects in monochromatic x-ray microdiffraction using tapered capillary optics

Tapered capillaries are frequently used as beam-concentrating optics in microbeam x-ray diffraction experiments. The beams exiting such devices are usually highly divergent and may possess nonuniform intensity distributions. In addition, their alignment poses some special challenges. In this article, the effects of these factors on the precision and accuracy of diffraction data are presented.

Topographic measurement of electromigration-induced stress gradients in aluminum conductor lines

We report a set of data on the evolution of stress in thin-film metallization wires during the transient region of electromigration. The excellent strain sensitivity of the x-ray microbeam topography technique allows real-time, spatially resolved measurements at the lowest currents reported to date (1.0×104–1.4×105 A/cm2). While the steady-state results agree qualitatively with the Blech’s stress gradient model [I. A. Blech, J. Appl. Phys. 47, 1203 (1976)], the threshold-length product calculated from our data is about 2–3 times smaller than previously reported values. Stress evolution during the transient state displays local fluctuations which cannot be attributed to experimental errors, indicating possible microstructural effects on local flux divergence even in the case of wide, nonbamboo wires.

A cost-effective method for minimizing the sphere-of-confusion error of x-ray microdiffractometers

Microdiffractometers are used to obtain x-ray diffraction data from regions that are tens of microns or less in size. If a microdiffractometer’s rotation circles do not share the same center, or if the feature of interest on a sample does not lie at the center of all rotations, the sample feature will, upon rotation of the diffractometer circles, precess through a finite volume known as the sphere of confusion (SoC). If the size of the beam used for diffraction analysis is smaller than the SoC diameter, the beam may actually move off the region of interest. In this article, we describe a new technique, based on x-ray fluorescence imaging and coordinate transforms, which can maintain the sample position to within ±6 μm over all rotations even when a commercial diffractometer is used as the base for the microdiffractometer system. In this scheme, a grid held in place on the specimen surface is mapped using fluorescent radiation at various sample tilts. The transformation matrices, which relate the grid coor...