Introduction to Research

Our current research program focuses on understanding, control, and applications of localized nucleation of nanostrcutures upon patterned surfaces. By coupling advances in experimental and modeling methods, we are advancing understanding of the concept of "templated assembly," the study of epitaxial nucleation as controlled by local surface variations. These advances provide the foundations for exploration and development of new nanostructured systems. Our current emphasis is on the epitaxial Ge(Si)/Si system, but we are extending the breadth of our discoveries through programs on templated assembly of metal oxide nanostructures and endothelial cells.

The Ga+ focused ion beam (FIB) is a highly versatile method for nanoscale control of surface structure and chemistry, allowing us to locally modify the surface according to surface topography, surface chemistry, surface strain fields, surface crystallinity, and surface reactivity. We have used this approach to demonstrate exquisite control of Ge nucleation on FIB-patterned Si(100), Figure 1.

These experiments have been performed in a unique UHV transmission electron microscope (TEM) in collaboration with Frances Ross’ group at IBM Yorktown Heights, which is equipped with digermane and disilane growth sources, such that Ge(Si) epitaxial growth may be performed during TEM imaging. Further, using MRSEC funding, we have equipped this system with an integrated Ga+ FIB column, such that FIB surface modification, Ge(Si) deposition, and TEM imaging may all be performed within the same inter-connected UHV system. We have shown that Ga+ doses as low as 1014 cm-2 at write speeds of 100 ms/feature, enable nanoscale control of Ge QD nucleation .

We have examined the effects of Ga+ dose and resulting topography upon the localization of Ge cluster nucleation, demonstrating that doses as low as 1014 ions cm-2 – which will not be expected to create even monolayer topography via sputtering – can effectively seed subsequent Ga+ nucleation. We have compared the efficiency of nucleation localization and rates during our standard CVD (where surface reactivity affects pyrolosis rates of the depositing species) and solid source deposition (from an effusion cell in an interconnected UHV chamber), showing that the resultant Ge cluster arrays are essentially identical, and that local variations in surface reactivity are thus not the dominant localization mechanism. We have also examined the role of surface chemical effects from implanted Ga. By chemically homogenizing the surface through deposition of a monolayer of Ga (from an effusion cell in an inter-connected UHV chamber) onto the as-implanted surface, we show that the resultant distribution of Ge clusters retains memory of the underlying implant pattern. This demonstrates that variations in surface chemistry are not the dominant mechanism in localizing nucleation. However, we do observe that cluster shapes and coherent-dislocated transition dimensions are different as a function of the surface Ga+, providing a route to controlling of cluster geometries and sizes.

A recent breakthrough has been the discovery that an unexpected nanoscale surface topography plays a key role in the localization of cluster nucleation. While our standard Ga+ ion dose of 1014 cm-2 is insufficient to directly create substantial surface topography, we observe that the post implant anneal that we perform to remove crystalline damage can produce a very subtle surface topography associated with each implanted feature. This initially comprises small annular surface depressions, of order 30 nm in diameter and 0.5 nm deep that, on further annealing, collapse to nano-pits c. 1 nm deep and 10 nm wide, near the center of each implant feature. Under annealing conditions where these depressions form (e.g. 550 o C for a few minutes), the Ge cluster localization is very apparent. Under annealing conditions where depressions do not form (i.e. at higher temperatures and/or longer annealing times) the majority of the Ge clusters do not nucleate on the implant sites.

Guided by these fundamental insights, we have been able to optimize Ga+ implantation, damage annealing, and deposition conditions to create arrays where every Ge cluster forms upon a FIB nucleation site, and the majority (> 95%) of the templated sites are occupied by clusters. This enables complex patterns of substantial complexity to be assembled. In Figure 2 , we demonstrate Ge cluster assembly in the symmetry of a quantum cellular automata (QCA) adder circuit. We believe that this demonstrates unprecedented control of epitaxial cluster nucleation.

the system; strong elastic and surface anisotropies, chemical interaction of the film and substrate for modeling wetting dynamics; epilayer deposition; bulk and surface diffusion (via an atomic mobility function that is localized in the diffuse interface region); substrate topographical patterning; and complex thin film morphologies such as faceting of quantum dot (QD) structures (via a polynomial function for dependence of surface energy upon orientation that captures low energy facets), step structures and topological transitions. Figure 3 shows a snapshot of cluster evolution during deposition of an epitaxial thin film. Guided growth occurs via the strain fields associated with a buried inclusion below the film/substrate interface, directly below the largest pyramid.

Other projects that explore the component effects of localization of cluster nucleation include: (i) Local/masked ion implantation to produce buried strain centers in Si (e.g., by formation of buried coherent metal silicide precipitates) with controlled magnitude and periodicity, to quantify the role of strain on Ge QD nucleation; (ii) Use of FIB tomographic reconstructions to explore 3-D correlations by propagation of strain fields in QD super-lattice structures, Figure 4, enabling us to track the size evolution of every QD through successive layers; (iii) Novel nanoscale electrochemical methods for creating complex 3-D surface topography, using both ultra-small machining electrodes and ultra-short charging pulses to localize the electrochemical reaction, and (iv) Development of new Ge-Ga and Si-Ga embedded atom potentials to enable atomistic modeling of adatom motion across Ga+ FIB- modified surfaces.

We have advanced these self-assembly mechanisms to multiple length scales, to enable the hierarchical structures necessary for application of our fundamental discoveries. By optimization of FIB surface modification, annealing and deposition parameters, we have demonstrated control of Ge QD nucleation with tens of nm precision, over fields of tens of square mm, with every QD nucleating on a FIB seeding site and more than 95% of the FIB sites occupied by QDs, at FIB write rates of 104 features/s. This demonstrates potential to scale to exploratory architectures of millions of features. A parallel route to hierarchical self assembly is our greatly enhanced understanding and control of self-assembling quantum dot molecules (QDMs), initially discovered in a separate NSF program (DMR-FRG 0075116), in G xSi1-x/Si(100). A QDM comprises a 4-fold set of {510} facetted islands bounding a {510} faceted pit. This provides a new component for our self-assembly toolset, and a geometry that lends directly to application to quantum cellular automata architectures. We have explained the observed size stability of these structures by experimental and modeling studies of their energetic and kinetic evolution. We have confirmed by photoluminescence measurements that carriers are confined to the dots comprising the molecules. We have also used topographic patterning to force QDMs into organized arrays over much larger lengths scales, Figure 5.

The MRSEC Center for Nanscopic Materials Design has greatly advanced the understanding and control of localized epitaxial cluster nucleation. Our IRG programs have generated advances relevant to the broad field of nanomaterials: (i) FIB tomographic methods with 3-D nanoscale structural/chemical resolution, (ii) Combined FIB/e-beam lithography for visual alignment to complex nanostructures (employed in preliminary attempts to realize operating semiconductor QCA cells using FIB and QDM architectures described above), (iii) Nanoscale electrochemical patterning using ultra-high resolution electrodes and ultra-short charging transients, enabling new multi-level topographic methods at spatial resolution c. 100 nm and possible new self-aligned etching / deposition schemes, (iv) Development of state-of-the-art 3-D continuum simulators for epitaxial surface evolution (c. 1000x faster than existing explicit methods) based on implicit time discretizations and multigrid methods.