William Huber

Superconformal Electrodeposition of Copper in 500–90 nm Features

Superconformal electrodeposition of copper in 500 nm deep trenches ranging from 500 to 90 nm in width has been demonstrated using an acid cupric sulfate electrolyte containing chloride (Cl), polyethylene glycol (PEG), and 3‐mercapto‐l‐propanesulfonate (MPSA). In contrast, similar experiments using either an additive‐free electrolyte, or an electrolyte containing the binary combinations Cl‐PEG, Cl‐MPSA, or simply benzotriazole (BTAH), resulted in the formation of a continuous void within the center of the trench. Void formation in the latter electrolytes is shown to be reduced through the geometrical leveling effect associated with conformal deposition in trenches or vias with sloping sidewalls. The slanted sidewalls also counterbalance the influence of the differential cupric ion concentration that develops within the trenches. Examination of the i-E deposition characteristics of the electrolytes reveals a hysteretic response associated with the Cl‐PEG‐MPSA electrolyte that can be usefully employed to monitor and explore additive efficacy and consumption. Likewise, resistivity measurements performed on corresponding blanket films can be used to quantify the extent of additive incorporation and its influence on microstructural evolution. The films deposited from the Cl‐PEG‐MPSA electrolyte exhibit spontaneous recrystallization at room temperature that results in a 23% drop in resistivity within a few hours of deposition.

Superconformal Electrodeposition of Copper

A model of superconformal electrodeposition is presented based on a local growth velocity that is proportional to coverage of a catalytic species at the metal/electrolyte interface. The catalyst accumulates at the interface through reaction with the electrolyte. More importantly, if the concentration of the catalyst precursor in the electrolyte is dilute, then surface coverage within small features can change far more rapidly due to changing interface area. In such a case, the catalyst effectively floats on the interface during deposition, with changes in coverage coupled to alterations in arc-length of the moving surface. The local coverage therefore increases during conformal growth on a concave surface, resulting in a corresponding increase in the local deposition rate. The opposite is true for a convex surface. The model is supported by experiments and simulations of superconformal copper deposition in 350-100 nm wide features. The model also has significant implications for understanding the influence of adsorbates on the evolution of surface roughness during electrodeposition.

A Simple Equation for Predicting Superconformal Electrodeposition in Submicrometer Trenches

We present a single variable first-order differential equation for predicting the occurrence of superconformal electrodeposition. The equation presumes that the dependence of deposition rate on surface coverage of the accelerator is known (e.g., derived from voltammetry experiments) on planar electrodes A simplified growth geometry, based on the recently proposed mechanism of curvature enhanced accelerator coverage, is used to permit simplification of the trench-filling problem. The resulting solution is shown to reduce computational time from hours to seconds, while yielding reasonably accurate predictions of the parameter values required for trench filling.

Excellent charge offset stability in a Si-based single-electron tunneling transistor

We have measured the long-term drift and the short-term 1/f noise in the charge offset Q0(t) in two Si-based single-electron tunneling transistors (SETTs). In contrast to metal-based SETTs, these devices show excellent charge stability, drifting by less than 0.01e over weeks. The short-term 1/f noise magnitude is similar to the metal-based devices, demonstrating that different mechanisms are responsible for the short-term noise versus the long-term drift. Finally, we show that, in addition to the excellent stability over time, it may be possible to make the devices more robust with respect to voltage-induced instability as well.

Why the long-term charge offset drift in Si single-electron tunneling transistors is much smaller (better) than in metal-based ones: Two-level fluctuator stability

A common observation in metal-based (specifically, those with AlOx tunnel junctions) single-electron tunneling (SET) devices is a time-dependent instability known as the long-term charge offset drift. This drift is not seen in Si-based devices. Our aim is to understand the difference between these, and ultimately to overcome the drift in the metal-based devices. A comprehensive set of measurements shows that (1) brief measurements over short periods of time can mask the underlying drift, (2) we have not found any reproducible technique to eliminate the drift, and (3) two-level fluctuators (TLFs) in the metal-based devices are not stable. In contrast, in the Si-based devices the charge offset drifts by less than 0.01e over many days, and the TLFs are stable. We also show charge noise measurements in a SET device over four decades of temperature. We present a model for the charge offset drift based on the observation of nonequilibrium heat evolution in glassy materials, and obtain a numerical estimate in go...

Using a high-value resistor in triangle comparisons of electrical standards

We propose an experiment with some advantages over other direct quantum metrology triangle comparisons. First, by using a cryogenic resistor that can be calibrated, the quantized Hall resistance (QHR) standard needs to be used only for short periods. Second, the experiment does not require a voltage detector. This eliminates one external source of noise and allows fast current reversals. Third, feedback that might contribute to excess noise and superconducting quantum interference device flux-jump behavior is also absent in the primary comparison system. This experiment could be run at higher currents and without supervision for extended periods of time, to benefit from statistical reduction of noise. We have developed a cryogenic current comparator for calibrating the cryogenic resistor directly against the QHR.

Excellent charge offset stability in Si-based SET transistors

We have obtained a large improvement in one aspect of the single-electron tunneling (SET) charge offset, Q/sub 0/, problem. The long-term drift of Q/sub 0/ makes it infeasible or impossible to integrate many devices. In contrast to metal-based SET transistors (SETTs), the long-term drift in Si-based SETTs appears to be at least one thousand times smaller (better).

Cu electrodeposition for on-chip interconnections

The electrochemical behavior of copper in copper sulfate -sulfuric acid, containing various combinations of NaCl, sodium 3-mercapto-1 propanesulfonate (MPSA), and polyethylene glycol (PEG) is examined. The i-E (current-potential) deposition characteristics of the electrolytes reveal a hysteretic response associated with the Cl-PEG-MPSA system that can be usefully employed to monitor and explore additive efficacy and consumption. Likewise, resistivity measurements performed on corresponding blanket films can be used to quantify the extent of additive incorporation and its influence on microstructural evolution. The films deposited from the Cl-PEG-MPSA electrolyte exhibit spontaneous recrystallization at room temperature that results in a 23% drop in resistivity within a few hours of deposition.

Microscope of glassy relaxation in femtogram samples: Charge offset drift in the single electron transistor

By measuring the long-term charge offset drift in single electron tunneling transistors, we have observed a transient relaxation after fabrication which is correlated with the presence of amorphous insulator. The temperature and time dependence of the transient relaxation are both in agreement with an extension of the standard model for two-level systems in glasses. This technique, which is sensitive to atomic scale motion in femtogram-sized samples, offers the possibility of a technique for investigation of glassy relaxation.

Using a high-value resistor in single-electron counting measurements

An experiment using a high-value cryogenic resistor is proposed, with the aim of improving the experimental link between three quantum electrical standards. This technique is new in its approach and does not require feedback from either a voltage detector or a current detector. Performance is limited by fundamental resistor and SQUID detector noise.