Microelectronic devices, such as semiconductor devices, are generally fabricated on and/or in wafers or workpieces. A typical wafer plating process involves depositing a seed layer onto the surface of the wafer via vapor deposition. The wafer is then moved into an electroplating processor where electric current is conducted through an electrolyte to the wafer, to apply a blanket layer or patterned layer of a metal or other conductive material onto the seed layer. Examples of conductive materials include permalloy, gold, silver, copper, and tin. Subsequent processing steps form components, contacts and/or conductive lines on the wafer.
In some electroplating processors, a current thief electrode, also referred to as an auxiliary cathode, is used to better control the plating thickness at the edge of the wafer and for control of the terminal effect on thin seed layers. The terminal effect for a given seed layer increases as the electrical conductivity of the electrolyte bath increases. Hence, a current thief electrode can be effectively used with thinner seed layers combined with high conductivity electrolyte baths. The use of thin seed layers is increasing common with redistribution layer (RDL) and wafer level packaging (WLP) plated wafers. For example, it is expected that RDL wafers may soon have copper seed layers as thin as 500 A-1000 A and copper bath conductivities of 470 mS/cm or higher.
In WLP processing, a relatively large amount of metal is plated onto each wafer. Consequently, in a WLP electrochemical processor having a current thief electrode, a large amount of metal will also be plated on the current thief electrode.
This metal must be deplated or otherwise removed from the current thief electrode at frequent intervals, with the processor removed from use during the deplating operation. Deplating the current thief electrode can also result in contamination particles in the electrolyte bath.
Damascene electroplating processors have used a current thief electrode, in the form of a platinum wire, inside of a membrane tube. The membrane tube holds a separate electrolyte (referred to as thiefolyte) having no metal (e.g., a 3% sulfuric acid and deionized water solution). The thief cathode reaction mostly evolves hydrogen rather than plating copper onto the wire. The hydrogen is swept out of the tube by the flowing thiefolyte. However, some metal does cross the membrane into the thiefolyte and plates onto the platinum wire (especially when using a lower conductivity bath). Consequently, the thiefolyte is only used once and flows to drain after passing through the membrane tube. The platinum wire is deplated after processing each wafer. However, under certain conditions using high thief current, it may be difficult In fully deplate the platinum wire.
The amp-minutes involved in processing RDL and WLP wafers can be 20 to 40 times higher than for damascene. As a result, the wire in a membrane tube thief electrode used in damascene electroplating may not suitable for electroplating RDL and WLP wafers, due to excessive metal plating onto the thief electrode wire, and excessive consumption of thiefolyte. Accordingly, engineering challenges remain in designing apparatus and methods for electroplating RDL and WLP wafers, and other applications, using a thief electrode.
In a first aspect, an electroplating processor has a vessel holding a first electrolyte or catholyte containing metal ions. A head has a wafer holder, with the head movable to position the wafer holder in the vessel. One or more anodes are in the vessel. A second electrolyte or isolyte in a second compartment is separated from the catholyte by a first membrane. A third electrolyte or ihiefolyte in a third compartment is separated from the isolyte by a second membrane. A current thief electrode is in the thiefolyte. The current thief electrode is connected to an auxiliary cathode and provides a current thieving function during electroplating. Build-up of metal on the current thief electrode is reduced or avoided via the membranes preventing metal ions from passing from the catholyte into the thiefolyte.
In the drawings, the same element number indicates the same element in each of the views.
Turning now in detail to the drawings, as shown in
Referring to
The current thief electrode assembly 92 may be used in a processor 20 having anodes 76 and 82 in the form of a wire-in-a-tube. A thief electrode wire 94 is provided in the thiefolyte channel 96 in the current thief electrode assembly 92. Virtual thief current channels 102 extend up through the vessel from the current thief electrode assembly 92 to a virtual thief position 99 near the top of the vessel, beyond the edge of the wafer 200.
Similarly, one or more second anodes 82 in an outer anode compartment are also provided in the form of an inert anode wire in a membrane tube. The anodes Flow diffusers 78 and 84 may be used, with the anode tubes on the outlet side of the diffusers. The diffusers may have tabs for holding the membrane tubes down against the floor of the anode compartment. During use, the catholyte chamber holds a liquid electrolyte, referred to as catholyte. Typically, a solution of sulfuric acid and deionized water, referred to as anolyte, circulates through the membrane tubes of the anodes 76 and 82. The circulating anolyte sweeps oxygen evolved off the inert anode wires within the tubes. The anolyte also provides a conductive path for the electric field from the inert anode wire to the catholyte.
Referring still to
The catholyte 202 in the channels 102 conducts the electric field created by the current thief electrode assembly 92 to the virtual thief position 99. In this way, the current thief electrode assembly 92 simulates having an annular thief electrode near the top of the vessel assembly 50.
In
Turning to
The isolyte 110 greatly reduces the amount of metal ions that are carried into the thiefolyte 104. In the case of a processor plating copper, because the isolyte 110 has a low pH and a very low copper concentration (as copper is only carried across the second membrane 100B) even a lower number of copper ions will be transported across the first membrane 100A and into the thiefolyte 104 touching the thief electrode wire 94. Thus, any plating onto the thief electrode wire will be very small. The catholyte solution for WLP has a low pH (high conductivity) and so the copper flow across the membrane separating the catholyte and the isolyte is low. In turn, the isolyte has both a low pH and a low copper concentration. These factors combine to yield an even lower flow of copper across the membrane separating the isolyte and the thiefolyte.
If the isolyte 110 is also the anolyte solution flowing through the membrane tubes of the anodes 76 and 84, some of the copper ions that get into the anolyte/isolation solution will pass through the anode membrane tubes and back into the catholyte 202. Furthermore, by greatly reducing the amount of copper transported into the thiefolyte 104, the thiefolyte 104 may be recirculated rather than used only once. Recirculating the thiefolyte 104 greatly reduces processing costs compared to using the thiefolyte only once as is done with damascene wafer processors. The small amount of copper that does make it to the thiefolyte 104 may plate onto the thief electrode wire 94, but only in small amounts that can be quickly deplated between wafers.
The fluid compartments illustrated in
As shown in
The apparatus and methods described provide a current thieving technique for plating WLP wafers, while overcoming the maintenance issue of copper plate-up on the thief electrode. This may be achieved by a two-membrane stack using cationic membranes and high conductivity (low pH) electrolytes. The copper containing catholyte is separated from a low-copper isolyte by a cationic membrane, which in turn is separated from the lower-copper thiefolyte by another cationic membrane. The thief electrode resides within the thiefolyte. The combination of chemistries and membranes resists migration of copper ions to the thief electrode.
This two-membrane design, with the thief electrode separated from the catholyte in the vessel by two membranes and two electrolytes, is suitable for preventing copper build on the thief electrode during long amp-minute wafer level packaging electroplating. The two separating electrolytes can be the same conductive fluid (i.e. acid and water). The two separating membranes can be cation or monovalent membranes. The separating isolyte and thiefolyte chambers can be formed as a stack with planar membranes, or the two membranes can be formed using co-axial tubular membranes with the inner rube membrane containing the thiefolyte and a wire thief electrode. The thief assembly mid -compartment can be the same electrolyte as the anolyte flowing over inert anodes within the process chamber.
Alternatively, a single membrane may be used to separate the catholyte from the thiefolyte. The catholyte contains copper but has a low pH. The thiefolyte is intended to have no copper. The membrane can be an anionic membrane that prevents copper ions from passing or a monovalent membrane that offers more resistance to Cu++ ions. In the single membrane design, the thief electrode is separated from the catholyte 202 by a single membrane, such as a flat or planar anionic membrane, and the thief electrode assembly has a single compartment. As used here, separated from means that the electrolytes on either side of a membrane are both touching the membrane, to allow the membrane to pass selected species as intended.
In
Conceptually, a centrally located thief acts circumferentially, beyond the edge of the wafer though a virtual anode channel. Since the thief current is relatively small compared to the anode currents, it is adequate to have a small, centrally located thief electrode (and its associated structure) rather than a thief electrode or assembly equal to or greater the circumference of the wafer as in currently used processor designs.
In a processor 20 without a paddle agitator, the virtual thief position or opening 99 may be below the wafer plane as shown in
Referring back to
Thus, novel apparatus and methods have been shown and described. Various changes and modifications may of course be made without departing from the spirit and scope of the invention. The invention, therefore, should not be limited, except to the following claims, and their equivalents.