Electrodeposition on a Resistive Patterned Wafer

This example models a cupplater reactor, where copper is electrodeposited on a patterned silicon wafer. It is also demonstrated how to make the deposited layer more uniform by the use of a current thief.

The deposition reaction occurs on a seed layer on an otherwise insulating wafer, with the effect that all current going into the metal layer is transported tangentially along the surface. As the deposition process progresses, and more metal is added to the surface, the tangential conductivity of the layer increases.

To cope with the very different geometric scales of the metal layer (tens of micrometers) and the electrolyte chamber (millimeters), which would otherwise pose problems in regard to mesh quality and problem size, the current conduction in the wafer metal layer is modeled using the Electrode, Shell interface, which uses a boundary formulation for the metal current conduction.

The model qualitatively reproduces the results of Purcar and others (Ref. 1).

Model Definition

The geometry is shown in Figure 1 and consists of one electrolyte domain. The top boundary is the anode. The bottom circular area represents the wafer surface, on top of which a metal seed layer has been deposited, followed by the application of a photoresist mask.

In the cell, the metal deposits on the patterned surface, which acts as cathode. The effect of the ring current thief is investigated by letting the surface be inactive as electrode in a first study, and then comparing the results to when letting also the current thief act as cathode. The area between the current thief and pattern is covered by a photoresist, and is inactive to the deposition reaction. A small edge segment of the current thief ring is used as current collector. The area outside the ring is insulating and does not conduct current in this model.

Figure 1: Model geometry.

The model is set up using two different interfaces: a Secondary Current Distribution interface, which solves for the electrolyte potential,

, and the deposited concentration, cdep, on the patterned wafer, and an Electrode, Shell interface, which solves for the electric potential on the metal layer on the wafer,

Secondary Current Distribution

The single domain is modeled as an Electrolyte domain. A constant conductivity of 20 S/m is used for the electrolyte.

The anode is assumed to a have a negligible polarization and an Electrolyte Current boundary condition, with a condition for the total current of 0.4 A, is used for the top boundary.

An Electrode Surface boundary node, with an added Dissolving-Depositing species, is used for the active cathode boundaries, with an electric potential for this electrode set to the electric potential of the Electrode, Shell interface. The kinetics are described by Butler-Volmer kinetics, and the thickness of the deposited layer is based on the molar mass and density of copper.

All other boundaries are isolated.

Electrode shell

An Electrode boundary is used for the inactive parts of the surface, where only current conduction occurs. A constant conductivity value, 5.6·107 S/m, is used for all parts of the surface. On the inactive parts, the metal layer thickness is equal to the seed layer thickness, 0.1 μm.

For the active parts of the wafer surface, the cathode, a Depositing Electrode boundary node is used. Here, the metal layer thickness is set to the seed layer thickness plus the electrode thickness change (due to the cathode depositing reaction), which is coupled to the Electrode Surface node in the Secondary Current Distribution interface.

The electrode reactions also give rise to a current source on the cathode boundaries, these are also coupled to the Electrode Surface node in the Secondary Current Distribution interface.

The current collector edge is grounded, whereas all other edges are isolated.

Studies

The problem is solved using a Time-dependent with Initialization, Fixed Geometry study type, simulating the deposition process for 600 s.

By using a study type for a fixed geometry, the change in geometry, which is expected to be in the range of micrometers, is not included in the variables that are solved for by the solver.

Two studies are performed. In the first, Study 1, only the pattern is used as active electrode on the wafer. In Study 2 both the ring current thief and the wafer pattern is used as cathodes.

Discussion

Figure 2 shows the electrolyte potential at the end of the Study 1, without the current thief. The potential in the outer parts of the patterned area is almost 40 mV higher than in the central parts.

Figure 2: Electrolyte currents and potential at the end of the simulation without a current thief.

Figure 3: Electric potential in the copper layer at the end of the simulation without a current thief.

Figure 3 shows the potential in the metal on the wafer at the end of the Study 1. Close to the current collector there is a significant potential drop. The potential is however fairly uniform for the active part of the wafer.

Figure 4: Electrode reaction current density at the end of the simulation without a current thief.

Figure 4 shows the electrode currents at the end of the Study 1. The reaction currents are significantly higher in the outer parts of the patterned area, especially in the corners. This is mainly an effect of the large differences in electrolyte potential seen in Figure 2.

Figure 5: Deposited thickness at the end of the simulation without a current thief. The surface is deformed in the z direction according to the layer thickness.

Figure 5 shows the resulting electrode thickness change at the end of Study 1. As a result of the higher electrode currents seen in Figure 4, the deposited thickness is nonuniform with higher thickness toward the outer rim of the pattern.

Figure 6: Electrolyte currents and potential at the end of the simulation with a current thief.

Figure 6 shows the electrolyte potential at the end of Study 2. Compared to Figure 2 the potential distribution is more uniform on the patterned surface.

Figure 7: Electric potential in the copper layer at the end of the simulation with a current thief.

Figure 7 shows the electric potential distribution at the end of Study 2 when using a current thief. The deposit build-up results in less steeper potential gradients close to the current collector. The potential over the patterned area is still fairly uniform.

Figure 8: Electrode reaction current density at the end of the simulation with a current thief.

Figure 8 shows the electrode current distribution at the end of Study 2. Compared Figure 4 to the electrode currents are more uniform on the patterned wafer. It can also be seen that the electrode currents densities are higher on the current thief than on the pattern, with a maximum at the current collector.

Figure 9: Deposited thickness at the end of the simulation without a current thief.

Figure 9 shows the deposited layer thickness at the end of Study 2. Compared to Figure 5 the deposited thickness is more uniform, all though the deposited thickness still is significantly higher in the corners of the pattern than in the central parts.

Figure 10: Comparison of deposited thicknesses, with current thief at t = 600 s, without current thief at t = 460 s.

Finally, the deposit uniformity effect of using a current thief is shown in Figure 10, where the deposit profile along the y-axis is compared for the two studies. The current thief results in a more uniform deposit, but also results in a longer deposition time due to the lower electrode current on the wafer pattern.

Reference

1. M. Purcar, B. Van den Bossche, L. Bortels, J. Deconninck, and G. Nelissen, “Three-Dimensional Current Density Distribution Simulations for a Resistive Patterned Wafer”, J. Electrochem.Soc., vol. 151, no. 9, pp. D78–D86, 2004.

Electrodeposition_Module/Tutorials/resistive_wafer

Modeling Instructions

From the menu, choose .

New

In the window, click