Precise wafer-to-wafer alignment accuracy is crucial to interconnecting circuits on different wafers in three dimensional integrated circuits. We discuss the use of fabricated structures on wafer surfaces to mechanically achieve higher alignment accuracy than can be achieved with our existing (baseline) alignment protocol. The keyed alignment structures rely on structures with tapered side-walls that can slide into each after two wafers are “pre-aligned” using our baseline alignment protocol. Results indicate that alignment accuracy is about a quarter micron, well below the one micron alignment accuracy obtained in our baseline alignment procedure using commercial state-of-the-art wafer alignment equipment. In addition to improving alignment, the alignment structures also hinder undesirable bonding-induced misalignment. The keyed alignment structures are also promising for nano-imprint lithography.

In this paper, we describe an experimental study of the nucleation and growth stages during Cu(TMVS)(hfac) sourced Cu CVD on TaN substrates. In particular, we have investigated the effects of water vapor as a co-reactant on evolving surface morphology. The results of short (less than 10 s) depositions without/with water vapor indicate that water vapor helps to reduce the incubation time and to enhance the nuclei formation, uniformity, and adhesion (based on AFM analysis). Introducing water vapor during only the initial stage of deposition results in lower roughnesses, larger grain sizes, and lower short-range roughnesses as compared to the films deposited without water vapor. From this study, we conclude that water vapor enhances Cu nucleation and that a relatively small amount of water vapor before or during the initial stage of deposition improves surface morphology in terms of roughness and grain size.

Conformality limitations and film composition variations inside features for films deposited at low pressures are explained using examples of studies which combine transport and reaction simulations of deposition processes with experimental work. In the first example, the use of film profile information to decide between two kinetic models for the low pressure chemical vapor deposition (LPCVD) of SiO2 from tetraethoxysilane (TEOS) is described. In the second example, a simple but useful model for the plasma enhanced chemical vapor deposition (PECVD) of silicon dioxide from TEOS/oxygen mixtures (PETEOS) is discussed. In the third example, composition profiles in sputter deposited (PVD) Ti-W films are explained in terms of titanium re-emission. Combined simulation and experimental studies of film profiles and composition profiles in features is a valuable tool in efforts to arrive at useful kinetic and transport models.

Experimental and simulation studies were conducted in an attempt to understand the effects of collimator life time on the Ti and TiN film growth rates and conformalities in sputter deposition processes. The Ti and TiN films were deposited with and without collimation. The hexagonal cells of the collimator used in this study have a 1:1 aspect ratio. A Monte Carlo based simulator was used to calculate the angular distributions of species exiting from a collimator cell and the percentage decrease in the rate of film growth as a function of the collimator life time. Then, a low pressure deposition process simulator, EVOLVE, was used to predict the conformalities of deposited films in contacts or vias, assuming that the films were uniformly deposited on the side-walls of collimator cells. We conclude that the loss in growth rate is largely due to the shrinkage in the cross sectional area of the collimator cell inlets. We arrive at this conclusion after comparing an estimated film thickness on the collimator side-walls with experimental measurements. With extended collimator usage, the predicted and experimental film profiles in contacts or vias show increasing bottom coverage and decreasing side-wall coverages.

Experimental and simulation studies were conducted in an attempt to understand the effects of collimator life time on the Ti and TiN film growth rates and conformalities in sputter deposition processes. The Ti and TiN films were deposited with and without collimation. The hexagonal cells of the collimator used in this study have a 1:1 aspect ratio. A Monte Carlo based simulator was used to calculate the angular distributions of species exiting from a collimator cell and the percentage decrease in the rate of film growth as a function of the collimator life time. Then, a low pressure deposition process simulator, EVOLVE, was used to predict the conformalities of deposited films in contacts or vias, assuming that the films were uniformly deposited on the side-walls of collimator cells. We conclude that the loss in growth rate is largely due to the shrinkage in the cross sectional area of the collimator cell inlets. We arrive at this conclusion after comparing an estimated film thickness on the collimator side-walls with experimental measurements. With extended collimator usage, the predicted and experimental film profiles in contacts or vias show increasing bottom coverage and decreasing side-wall coverages.

The growth morphology of Ag on GaAs (110) surface, at low coverage, is investigated with Monte Carlo simulations using a solid-on-solid model. Experimentally Ag deposited at room temperature forms isotropie islands and forms islands elongated along the <110> direction when deposited at 250°C. The preliminary results from our Monte Carlo simulations indicate that anisotropy in the surface migration and near-neighbor interactions are the main factors for the elongated islands observed for depositions performed at 250°C.

Simulation of the physical vapor deposition (PVD) of titanium nitride (TiN) in micronscale contacts was performed using EVOLVE, a physically-based model of non-continuum mass transport. Simulations of uncollimated TiN PVD using a sub-unity sticking coefficient of 0.6 provide accurate model predictions of experimental data. Using this value in simulations of PVD with a 1:1 aspect ratio collimator results in a significant underprediction of titanium nitride deposition in the feature bottom in comparison to the experimental results. This under-prediction was compensated for empirically by adjusting a “beaming” coefficient to produce good model predictions of the experimental data.

The characteristic step coverage behavior which a given LPCVD process exhibits depends on the nature of the controlling gas phase and/or surface chemical reactions. Physically-based ballistic transport and reaction film profile evolution simulation has provided a structure wherein the origins of step coverage limitations can be understood in the context of the interaction of transport and the controlling chemistry. Based on comparisons of the simulations to literature and in-house experimental data, we have categorized LPCVD mechanisms into one of three types. In heterogeneous deposition, conformal step coverage can usually be found under at least some process conditions. Step coverage typically degrades with increasing deposition temperature. In homogeneous precursor-mediateddeposition, a reactive intermediate is formed in the gas-phase above the wafer surface, resulting in poor to moderate step coverage. Step coverage may or may not degrade with increasing temperature. In byproduct-inhibited deposition, a gas-phase byproduct generated via a surface reaction readsorbs on the growing film surface and slows the deposition rate, yielding a poor to moderate, relatively temperature-insensitive step coverage. Poor step coverage is manifested in a marked film thickness discontinuity at the feature mouth, with a relatively uniform film down the feature sidewalls.

Diffusion-reaction analysis of a two step process in which a tungsten silicide barrier layer is deposited in a rectangular trench by low pressure dichlorosilane reduction of tungsten hexafluoride followed by a complete tungsten fill by low pressure hydrogen reduction of tungsten hexafluoride reveals that high step coverage and high deposition rate can be readily achieved with logical selection of process parameters. This fact, coupled with the potential for accomplishing these deposition steps in the same single wafer reactor, suggests that this two step process may offer a high throughput alternative to blanket tungsten deposition by silane reduction.

The deposition rate of silicon dioxide from TEOS/02 capacitively-coupled plasmas increases with increasing applied rf power, increasing total pressure and decreasing wafer temperature. These dependences can be explained by a mechanism in which deposition occurs through both an ionassisted and an oxygen atom initiated pathway. The ion-induced deposition rate increases with increasing rf power density and decreasing total pressure, and is independent of wafer temperature. Assuming a well-mixed plasma, and using literature values for reaction rate parameters and rate forms for oxygen plasma reactions, the rate of atom-induced deposition was found to be independent of temperature and TEOS concentration and directly proportional to oxygen atom concentration. The model suggests that the apparent negative activation energy for deposition results from competition between deposition and thermally-activated atomic oxygen surface recombination. The derived deposition rate form is consistent with an Eley-Rideal mechanism in which the rate determining step for film deposition is oxidative attack of adsorbed TEOS or TEOS fragments by a activated oxygen.

A modified line-of-sight model for transport and deposition during LPCVD is used to predict step coverage and film composition uniformity of tungsten silicide barrier layers. Predictions are compared with experimental results for 2 μim wide by 6 μm deep trenches with barrier layers of 0.2 μm nominal thickness. Model predictions are in quantitative agreement with those of a diffusion-reaction model and are in qualitative agreement with experiment.

A multicomponent mathematical model incorporating simultaneous Knudsen diffusion and heterogeneous reaction in a feature of arbitrary geometry has been developed to quantitatively predict step coverage in LPCVD. The single component model reveals that for a given CVD chemistry and feature geometry, deposition uniformity is controlled by a single parameter called the step coverage modulus. In multicomponent systems, each additional reactant partial pressure dependence in the deposition rate expression introduces an additional parameter equal to the ratio of the initial reactant partial pressures. Manipulation of the instantaneous step coverage modulus, through time variations of temperature and/or pressure results in higher average film growth rates without incurring a penalty in step coverage over patterned regions of the wafer. For deposition of SiO2 from TEOS, the step coverage modulus is varied through a time dependent temperature path. The path is chosen by constraining the deposition rate in the feature such that it varies by less than a few percent from top to bottom. For deposition of SiO2 from SiH4/O2, silane partial pressure is varied since the reaction rate is relatively insensitive to temperature. In each case. time savings of approximately 50% are achieved.

The distribution of nickel and copper between crystallites supported on high surface area silica (Cabosil HS5) has been studied using a Philips 400 series transmission electron microscope equipped with super twin lenses, a field emission gun and EDAX energy dispersive X-ray spectroscope. The same Ni-Cu catalysts have previously been characterized and used in cyclopropane hydrogenolysis kinetics. The good intercrystallite homogeneity found in a well reduced sample having 31% nominal copper validates a critical assumption made in these previous studies. Limited data on unreduced and partially reduced samples also highlight the need to carefully control the degree of reduction in kinetic and characterization studies.