Accurate pattern registration for integrated circuit tomography

Abstract

Erratum: "Explicit pure-state density operator structure for quantum tomography" [As part of an effort to develop high resolution microtomography for engineered structures, a two-level copper integrated circuit interconnect was imaged using 1.83 keV x rays at 14 angles employing a full-field Fresnel zone plate microscope. A major requirement for high resolution microtomography is the accurate registration of the reference axes in each of the many views needed for a reconstruction. A reconstruction with 100 nm resolution would require registration accuracy of 30 nm or better. This work demonstrates that even images that have strong interference fringes can be used to obtain accurate fiducials through the use of Radon transforms. We show that we are able to locate the coordinates of the rectilinear circuit patterns to 28 nm. The procedure is validated by agreement between an x-ray parallax measurement of 1.41Ϯ0.17 m and a measurement of 1.58 Ϯ0.08 m from a scanning electron microscope image of a cross section.

Parallax measurements of integrated circuit interconnects using a scanning transmission electron microscope

“…A sample with the same design parameters was analyzed using a cross section with a scanning electron microscope as well as using parallax analysis with an x-ray microscope. 5 The thicknesses were reported to be 1.58 m Ϯ0.08 m and 1.41 mϮ0.17 m with these two techniques, respectively, putting the STEM measurement in marginal disagreement and agreement with the two previous measurements.…”

“…3, was a two-layer copper sample in a low dielectric constant polymer matrix, taken from the same batch as a sample some of us used in x-ray microscopy. 5 The second sample, called the ''fuse bay'', shown in the bottom of Fig. 3, was a two-layer copper sample in a silica matrix.…”

“…4,5,9,13 The principal differences are: ͑1͒ here, we assume the normal of the wiring plane of the sample is aligned with the beam for some tilt angle, but previously we did not make this assumption and ͑2͒ previously, we assumed the rotation axis of the sample was aligned with the y direction but here, we allow the rotation axis of the sample to be in a general direction orthogonal to the beam direction.…”

“…A first study of an integrated circuit interconnect with a fieldimaging microscope was limited to a parallax analysis due to diffraction. 5 However, more recently an incoherent x-ray illumination has permitted the rapid acquisition of x-ray images of integrated circuit interconnects, at least for two-layer circuits. 6 The use of transmission electron microscopy for failure analysis in the semiconductor industry has traditionally been applied to samples which are typically 0.2 m thick.…”

Submicron imaging of buried integrated circuit structures using scanning confocal electron microscopy

“…A parallax analysis of the tilt series was performed according to the techniques outlined in Ref. 10. The alignment portion of the analysis yielded orientation angle values of , , and 0 to be 63.234°, Ϫ4.677°, and Ϫ2.392°, respectively.…”