Roger L. Alvis

Observation of diamond nanocrystals in carbon films deposited during ion‐assisted microwave plasma nucleation pretreatments

Diamond nanocrystals were observed in carbon films deposited by ion‐assisted microwave plasma pretreatments for diamond thin films. Both selected‐area electron diffraction and x‐ray diffraction confirm the presence of diamond in the bias‐deposited carbon films despite micro‐Raman spectra and scanning electron micrographs which do not provide conclusive evidence of this carbon phase. These analyses indicate that carbon deposited by the ion‐assisted nucleation process consists of a high density of nanocrystalline diamond nuclei in a nondiamond carbon matrix. Subsequent growth of these nuclei using unbiased microwave plasma chemical‐vapor deposition leads to the formation of relatively smooth and thin continuous polycrystalline diamond films with small grain sizes.

The influence of capping layer type on cobalt salicide formation in films and narrow lines

The effect of capping layer (Ti vs TiN) on CoSi formation and CoSi 2 sheet resistance has been investigated. Resistance measurements and XTEM analysis have been used to show that the Ti cap lowers the activation energy for CoSi formation by gettering the O 2 from the RTA (rapid thermal anneal) ambient and eliminating the formation of SiO 2 between the growing CoSi and the Co. The sheet resistance of cobalt silicide formed from Co/Ti and Co/TiN bilayers on poly- Si lines was measured as a function of linewidth and RTA temperature. With a Ti cap, the sheet resistance is low and independent of temperature, and the RTA process window is large.

Atomic force microscopy for cross section inspection and metrology

Images of integrated circuit cross sections may be acquired with the atomic force microscope (AFM) by introducing material‐dependent topography through a series of selective etches. AFM images of a fully processed complementary metal–oxide–semiconductor inverter structure show excellent qualitative agreement with high resolution scanning electron microscope (SEM) images. Measurements of layer thicknesses and lateral dimensions, however, do not precisely correlate. These discrepancies are attributed to tip–sample convolution due to the finite cone angle and rounding of the probe. We describe a one‐dimensional computer simulator that models the nonlinear geometrical interaction between a tip and sample. Simulation results are used to determine the tip shape from an AFM image of a feature of known dimensions. The tip influence can be subsequently deconvolved from a cross section AFM image, generating a more faithful reflection of the surface topography. We demonstrate that this scheme yields measurements tha...

Physical characterization of two‐dimensional doping profiles for process modeling

Physical characterization of doping profiles in two dimensions holds great promise for both high quality analysis of specific structures and for general physical model verification. This latter activity enables the calibration of process simulators and could lead to accurate predictive simulation of modern integrated circuit devices. We used both one‐ and two‐dimensional analytical techniques [secondary‐ion‐mass spectroscopy (SIMS) and transmission electron microscopy (TEM)] to quantitatively characterize implanted and rapid‐thermal‐annealed dopant profiles at a polysilicon gate edge. The samples were given self‐aligned arsenic implants of 1×1015 ions/cm2 at 35 and 120 keV and at 0° and 20° angles of incidence. The implant was followed by a 30 s/1000 °C rapid thermal anneal. SIMS profiles were used to calibrate 1D simulations and the TEM micrographs in the 1D regions far from the mask edge. Quantitative TEM micrographs near the gate edge were then compared with two‐dimensional simulations of final doping ...

Junction metrology by cross‐sectional atomic force microscopy

In this article we address practical aspects of the extension of cross‐sectional atomic force microscopy to junction metrology. Examining an arsenic (As+) ion implantation split lot, four specimens were epoxied into a single stack. Thus, each sample was exposed to the same preparation conditions. Using a noncontact mode of image acquisition, we verified that a 120 keV implantation resulted in deeper doping profiles than a 35 keV implant. A shift of the source and drain junctions with respect to the gate was also observed as the angle of implant was changed. Correlating the measured junction depths of each sample with secondary‐ion‐mass spectroscopy profiles, we determined that the detection limit of the junction etch was in the range of 1017–1018 cm−3. The absolute accuracy of this technique was partially limited by tip interaction with the large topography introduced by etching.

The Effect of Si ON TiAl3 Formation EV Ti/Al Alloy Bilayers

Metallization lines in advanced integrated circuits are often fabricated from sputter-deposited Ti/Al layers. It is well known that the Ti/Al react above ∼350°C to form T1AI3 with a rate that is dependent on anneal temperature and the alloying content of Cu and Si in the Al. In the present work, the thickness of TiAl 3 formed during annealing at 430°C has been determined from the measurement of the sheet resistance of Ti/Al-.5%Cu and Ti/Al-.5%Cu-l%Si bilayers. Analyses of the reacted structures were performed by cross-section transmission electron microscopy and Auger depth profiling. We find that the Si exhibits a greater retardation effect on the Ti/Al reaction than does Cu, as shown previously. Kinetic analysis for the Ti/AlCuSi reaction shows that the TiAl 3 formation rate is a function of the Ti/AlCuSi thickness ratio. We propose that this effect is due to several mechanisms which involve diffusion and incorporation of Si into the growing TiAl 3 layer.

Texture Analysis of Al and Cu Metallization Materials Using Orientation Imaging Microscopy

X-ray diffraction (XRD) is generally used to measure crystallographic texture, but complete analysis is rarely performed. Orientation imaging microscopy (OIM) gives similar data by electron diffraction and has potential for widespread use since it may be added to any scanning electron microscope. Analysis software has been developed to reduce QIM results into a form which may be compared directly to XRD results. OIM texture parameters measured from Al films showed good quantitative agreement with XRD parameters, however the texture in Cu films is more complex and difficult to compare. Although subject to certain limitations, OIM has several advantages over XRD including more efficient data collection, better sensitivity to minor texture components, and more flexible data analysis.