Facile Optimization of Combinatorial Sputtering Processes with Arbitrary Numbers of Components for Targeted Compositions

TL;DR

This work addresses the challenge of tuning multi-component sputtering processes to reach targeted film compositions. It introduces a calibration-based workflow that uses WDXRF composition mapping to predict deposition rates as functions of target power for each component and to interpolate an optimal power set for arbitrary component counts. The method requires at least two calibration depositions and yields a predictive power set that achieves the desired stoichiometry at specified wafer coordinates, demonstrated on Cr-Fe-Mo-Nb-Ta with equiatomic composition near wafer center, and validated by EDS cross-tech verification with WDXRF maps. This approach enables rapid, high-throughput exploration of complex alloy libraries by converting guess-and-check into a calculable optimization, with broad applicability to metallic thin films and beyond.

Abstract

Combinatorial sputtering is a physical vapor deposition method that enables the high-throughput synthesis of compositionally varied thin films. Using this technique, the effects of stoichiometry on specific properties of alloy thin films with analog composition gradients can be mapped using high-throughput characterization. To obtain specific stoichiometries, such as those desired for an equiatomic, intermetallic, or doped compounds, the sputter power of each target must be simultaneously tuned to optimize the deposition rate of each component. This optimization problem increases in complexity with the number of components, which commonly leads to iterative guess-and-check processing and can limit the intrinsic high-throughput advantages of this synthesis method. To circumvent this challenge, this work introduces a composition optimization procedure that enables the facile synthesis of sputtered combinatorial films with targeted compositions. This procedure leverages the expeditious mapping of composition using wavelength dispersive x-ray fluorescence and is capable of optimizing processing for an arbitrary number of components. As a demonstration, this method is leveraged to sputter a combinatorial Cr$_{v}$Fe$_{w}$Mo$_{x}$Nb$_{y}$Ta$_{z}$ film with an equiatomic composition near the wafer center.

Figures (9)

• Figure 1: (a) Normalized areal mass density profile of a 2.6 W cm$^{-2}$ Mo deposition on a 6-inch diameter silicon wafer measured using WDXRF. (b) Two-dimensional Gaussian profile optimized using the areal mass density profile shown in panel (a). (c) Percent error difference between the measured and fit profiles in panels (a) and (b), respectively. In each panel, warm colors correspond to values with higher magnitudes, while cooler colors correspond to values with lower magnitudes, as detailed by the color bars above each respective plot and labeled contour lines. The black dotted circle in panels (a) and (c) represents the extent of the 6-inch diameter Si wafer.
• Figure 2: Fit areal mass density amplitude rate of individually deposited Mo (red) and Cr (blue) versus deposition power at $x$ = $x_0$, $y$ = $y_0$. Linear fits to each Mo and Cr data set are shown as black dashed and dot-dashed lines, respectively, the ${R}$$^2$ values and slopes of which are annotated to the left and right. Error bars come from least squares amplitude fitting, and are $\leq$ 1% for each data point.
• Figure 3: (a) Areal mass density ratio between Cr and Mo of wafer codeposited with 3.9 W cm$^{-2}$ applied power density across both targets. Percent error between each individual (b) Cr and (c) Mo species in the codeposited wafer and associated 3.9 W cm$^{-2}$ calibration deposition. In each panel, warm colors correspond to values with higher magnitudes, while cooler colors correspond to values with lower magnitudes, as detailed by the color bars above each respective plot and labeled contour lines. The black dotted circle in all three panels represents the extent of the 6-inch dimater Si wafer in the codeposition.
• Figure 4: Fit areal mass density amplitude rate of co-deposited Ta (orange), Mo (red), Nb (purple), Cr (blue), and Fe (green) versus deposition power. Linear fits to each data set are shown as black dashed and dotted lines (see legend) and associated $R$$^2$ values are annotated above or below each corresponding line. Error bars come from least squares fitting of amplitude, and are $\leq$ 2% for each data point.
• Figure 5: Relative RMS deviation maps for a (a) simulated and (b) deposited Cr$_{v}$Fe$_{w}$Mo$_{x}$Nb$_{y}$Ta$_{z}$ film synthesized using powers optimized to obtain an equiatomic composition at the center position. The white star near the center of the map in panel (b) corresponds to the position with the lowest relative $\sigma$ value. In all panels, warm and cool colors indicate larger and smaller values, respectively, as detailed by each corresponding color bar. The black dotted circle in both panels represents the extent of the 6-inch dimater Si wafer in the codeposition.