May 22, 2007

2007 Joint NSLS/CFN Users’ Meeting Workshop

Applications of Synchrotron-Based Methods to Microelectronics Materials

On May 23, as part of the Joint NSLS/CFN Users’ Meeting, a workshop was held on “Applications of Synchrotron-Based Methods to Microelectronics Materials.” Its purpose was two-fold: to demonstrate how a variety of synchrotron techniques can be applied to this field, and to survey a wide range of topics in microelectronics materials research that have been explored using x-ray and UV synchrotron radiation. It is hoped that the attendees of this workshop will be inspired to use synchrotron methods to a greater extent in their work and that the microelectronics community will think about what capabilities they would like to see in NSLS and NSLS-II beamlines.

The morning session consisted of four talks that illustrated how synchrotron techniques can be used in studies that range from pure to applied research. Simone Raoux, of IBM Almaden Research Center, gave a comprehensive review of several years’ work using time-resolved diffraction, resistivity, and light scattering to characterize materials for phase-change, solid-state memory. Many characteristics are required for good “universal” phase-change memory, including a large change in resistance when transforming between amorphous (reset) and crystalline (set) states and a desirable crystallization temperature. Dr. Raoux showed results on doping experiments to confirm predictions of glass transition temperature modeling, scaling effects (in film thickness and nanodot diameter) on crystallization temperature, and solution-based processing of two-component precursors to tune crystallization temperature.

Participants of the “Applications of Synchrotron-Based Methods to Microelectronics Materials” workshop.

Joe Woicik of the National Institute of Standards and Technology (NIST) described elegant experiments using a combination of high-resolution x-ray diffraction (XRD), near-edge x-ray absorption fine structure (NEXAFS), and first-principles density functional theory calculations to understand the structural distortion in epitaxial SrTiO₃/Si(001) that causes it to become ferroelectric at room temperature. Compressive strain in the thin (~5 ML) film coherently grown on clean Si(001) causes a tetragonal distortion where the out-of-plane lattice constant far exceeds predictions of the bulk elastic constants, and an off-center displacement of the central Ti atom. Calculations of the entire film/substrate structure confirm an energetically favorable interfacial-surface-charge structure that allows the ferroelectric polarization. This engineered strain epitaxy system demonstrates the possibility of producing ferroelectric SrTiO₃ devices on Si.

The last two talks of the morning were descriptions of studies of metal silicide texture and formation. Synchrotron-based, pole-figure measurements were used by Christophe Detavernier of the University of Gent, Belgium, to understand texture in thin metal silicide films. Patterns of tilted circles and arcs in pole figure plots led to the discovery of axiotaxy, a new type of texture where certain lattice planes of the film preferentially align themselves with a set of planes in the substrate (in this case the 220 planes in silicon), resulting in a 1-D-like epitaxy. For many silicide and germanide films on silicon, axiotaxial alignment is the main type of preferred grain orientation and can drastically effect agglomeration. The talk ended with a discussion of the pros and cons of using XRD, electron backscatter diffraction (EBSD), and transmission electron microscopes (TEMs) to measure thin film texture.

Silicides were also the topic of interest for Jorge Kittl of Texas Instruments at IMEC. He presented a vast amount of work detailing the sequence of phase formation of nickel silicides in a series of Ni/poly-Si/SiO₂/Si samples (modeling FUSI gate stacks) having varying ratios of Ni film to poly-silicon thicknesses in order to tune composition. In addition to determining competing phases and their sequence of appearance, kinetics was studied both by varying RTA ramp rate (using Kissinger analysis) and by isothermal ripening (using modified Avrami analysis). Isothermal analysis showed that Ni₂Si is diffusion controlled 1-D growth, Ni₃₁Si₁₂ and Ni₃Si₂ follow 2-D growth, and Ni₃Si has a long incubation time followed by 2-D growth. Ni grown on poly-Si narrow lines behaves differently and produces Ni-rich phases due to lower availability of Si reactant.

The first afternoon session focused on the use of micro and nanobeam techniques and was held jointly with Workshop 8, “Making and Using Nanobeams.” Leading off, Michael Feser, of Xradia Inc., described his company’s full-field phase-contrast x-ray transmission microscope and computed microtomography system, “nanoXCT,” recently installed at the Taiwan synchrotron, Stanford Synchrotron Radiation Laboratory, and the Advanced Photon Source. A full 3-D tomograph with sub-30nm resolution can be obtained in less than a minute using the high brilliance available at a synchrotron. In the nanoprobe scanning mode (at APS ID-26), elemental (fluorescence) mapping and nanobeam diffraction are also possible, with strain resolution on the order of 10^-4 to 10^-5. The system operates from 3 to 20 keV.

Conal Murray, of IBM T.J. Watson Research Center, reported on the use of microbeam diffraction to study the response of thin silicon films and substrates to both epitaxially deposited and imbedded strained features that are used to engineer strain into transistor carrier channels, thereby enhancing electron or hole mobility. The strain relaxation due to free edges of narrow (1 to 20 micron) SiGe lines on Si(001) was measured and modeled using a variety of analytical models and numerical simulations. It was found that the shear-lag model best describes the strain in this geometry. The strain induced in the substrate was mapped in real space using x-ray scanning topography and simulated using dynamical diffraction theory and the edge-force model. Embedded shallow trench isolation features in silicon-on-insulator films were found to induce the greatest extent of strain, up to 25 times the film thickness.

As an interesting departure from complementary metal–oxide–semiconductor (CMOS) systems, Tonio Buonassissi, recently from Evergreen Solar and currently at MIT, lectured on studies of metal impurity identification and distribution in solar photovoltaic materials. A combination of techniques (x-ray absorption microspectroscopy, x-ray fluorescence microscopy, and x-ray beam-induced current) allowed him to identify metal species (usually silicides), map distributions and particle sizes, and evaluate the performance of multigrain Si thin film materials. Particles such as Fe silicides that are small and evenly distributed affect performance to a much greater extent than the same impurity content concentrated into large particles. Results have led to creative defect engineering by altering metal precipitate formation behavior and promoting formation of deleterious species in regions that don’t affect device performance.

The final two talks of the afternoon addressed the newly developing technology of high-k gate dielectrics. These films allow for a thicker dielectric layer under the gate while still maintaining the electrical “equivalent oxide thickness” of ultra-thin silicon oxides and oxynitrides. Patrick Lysaght, of Sematech, described how he uses a combination of synchrotron techniques as well as small-angle neutron scattering (SANS), electron energy-loss spectroscopy (EELS), and TEM to understand many aspects of processing of Hf oxide and silicate thin films. These included effects of layer thickness, N2 vs. NH3 predeposition nitridation, annealing, diffusion of Al₂O₃ and La₂O₃ cap layers, and Ru electrodes. Two clever uses of synchrotron methods are noted in particular. First, variable kinetic energy and variable angle x-ray photoelectron spectroscopy (XPS) were used to measure the change in the bonding environment going from the surface of HfO₂ films through to the HfO_x/SiO_x interface, where O-deficient SiO2 was found. Second, for studies of amorphous to crystalline transitions in HfO₂ as a function of temperature and thickness, calculated EXAFS Fourier transform radial plots for three known crystalline phases were used to identify crystallinity in films that showed no evidence in XRD data.

The final talk of the day was given by Martin Green, of NIST, who reported on the use of X-ray Reflectometry (XRR) and Grazing-Incidence Small-Angle X-ray Scattering (GISAXS) to study the nucleation, growth, and morphology of atomic-layer-deposited HfO₂ high-k gate dielectric layers. Minimal thickness and fully cohesive, pore-free films are required for maximum gate capacitance in advanced CMOS transistors. These techniques were used to determine formation, growth and coalescence of nuclei as well as film porosity and roughness. HfO₂ was grown from several precursor materials on two types of Si substrates. Films on H-terminated Si were up to 50% porous and rough, but those grown on chemically oxidized Si were much denser. Film nuclei were 20 -25 Å and coalesced after a number of growth cycles, with the same feature size still persisting in the scattering signal, indicating that the films consisted of densely packed (5% porosity) monodisperse crystallites. The chemically oxidized substrate has many more nucleation sites for HfO₂ formation.

The workshop finished with a short discussion about the importance of the microelectronics materials community banding together to determine and vocalize its instrumentation needs at the NSLS and NSLS-II. Attendees were also invited to tour the NSLS after the session ended.

FOR MORE INFORMATION:
Jean Jordan-Sweet
IBM Research Division
Email: jlj@bnl.gov

Christian Lavoieb
IBM Research Division
Email: clavoi@us.ibm.com