US 20040065540 A1
A treating head having a treating surface and a substrate treatment surface define a thin fluid gap that is filled with reactant liquid to form a thin liquid layer on the substrate for conducting a liquid chemical reaction treatment or other liquid treatment of the substrate. The thin liquid layer has a volume in a range of about from 50 ml to 500 ml. Preferably, the chemical composition, temperature, and other properties of liquid in the thin liquid layer are dynamically variable.
1. An apparatus for thin-liquid-layer treatment of a surface of an integrated circuit substrate, comprising:
a substrate holder;
a treating head proximate to the substrate holder, the treating head including a head surface that forms a thin fluid gap between the head surface and a substrate treatment surface when a substrate is present in the substrate holder; and
a liquid inlet tube for flowing liquid into a thin fluid gap.
2. An apparatus as in
3. An apparatus as in
4. An apparatus as in
5. An apparatus as in
6. An apparatus as in
7. An apparatus as in
8. An apparatus as in
9. An apparatus as in
10. An apparatus as in
11. An apparatus as in
12. An apparatus as in
13. An apparatus as in
14. An apparatus as in
a manifold cavity disposed in the treating head;
a manifold inlet for providing fluidic communication between a liquid source and the manifold cavity; and
a plurality of liquid inlet tubes integral with the treating head for flowing liquid from the manifold cavity into a fluid gap when a substrate is present in the substrate holder.
15. An apparatus as in
16. An apparatus as in
17. An apparatus as in
a treating liquid source;
a containment chamber containing the substrate holder and having an outlet drain; and
a recycling tube between the outlet drain and the treating liquid source.
18. An apparatus as in
a containment chamber containing the substrate holder; and
a liquid diversion system wherein the liquid diversion system includes a collection trough, and the collection trough is disposed in a containment chamber substantially radially outwards from the substrate holder.
19. An apparatus as in
20. An apparatus as in
21. An apparatus as in
22. An apparatus as in
23. An apparatus as in
a liquid source tube from a treating liquid source;
a recirculation tube for recirculating treating liquid from a liquid source tube back to the treating liquid source; and
a liquid cooler for cooling recirculating treating liquid.
24. An apparatus as in
25. An apparatus as in
a plurality of treating liquid sources including a first treating liquid source and a second treating liquid source; and
a liquid mixer, the mixer being located proximate to the treating head, for mixing a first treating liquid from the first treating liquid source and a second treating liquid from the second treating liquid source.
26. An apparatus as in
27. An apparatus as in
28. An apparatus as in
29. An apparatus as in
30. An apparatus as in
a magnetic source for creating a magnetic field in a thin fluid gap; and
a magnetic sensor for detecting an amount of metal present on a substrate present in the substrate holder.
31. An apparatus as in
32. An apparatus as in
33. An apparatus as in
34. An apparatus as in
35. An apparatus as in
a light source; and
an optical sensor for measuring an optical property related to an amount of material deposited on a substrate treatment surface.
36. An apparatus as in claim-35 wherein said optical property is selected from a group consisting of optical reflectivity, optical transmittance, and optical spectrum.
37. A method of liquid treatment of a surface of an integrated circuit substrate, comprising:
placing an integrated circuit substrate having a treatment surface in a substrate holder;
disposing a treating head having a head surface proximate to the treatment surface, the head surface and the treatment surface thereby defining a thin fluid gap; and
flowing liquid into the thin fluid gap to form a thin liquid layer.
38. A method as in
39. A method as in
40. A method as in
41. A method as in
42. A method as in
43. A method as in
44. A method as in
45. A method as in
46. A method as in
47. A method as in
48. A method as in
49. A method as in
50. A method as in
51. A method as in
52. A method as in
flowing liquid into the fluid gap during a first period of time; and then substantially ceasing flowing liquid into the fluid gap during a second period of time.
53. A method as in
54. A method as in
55. A method as in
56. A method as in
flowing a first treating liquid into a mixer;
flowing a second treating liquid into the mixer to form a mixed liquid with the first treating liquid; and
then flowing the mixed liquid into the fluid gap.
57. A method as in
varying a property of the mixed liquid upstream of the fluid gap to form a second mixed liquid; and
then flowing the second mixed liquid into the fluid gap.
58. A method as in
59. A method as in
60. A method as in
61. A method as in
62. A method as in
63. A method as in
64. A method as in
heating treatment liquid upstream of the fluid gap;
diverting a recirculation-portion of the treatment liquid instead of flowing the recirculation-portion into the fluid gap;
cooling the recirculation-portion of treatment liquid; and
then flowing the recirculation-portion of treatment liquid to a treatment liquid source.
65. A method as in
66. A method as in
67. A method as in
68. A method as in
69. A method as in
70. A method as in
71. A method as in
72. A method as in
73. A method as in
74. A method as in
75. A method as in
76. A method as in
77. A method as in
flowing a nucleation solution into the fluid gap; and
then flowing a growth solution into the fluid gap.
78. A method as in
79. A method as in
80. A method of depositing a foreign-metal layer onto a base-metal material on a treatment surface of an integrated circuit substrate using a thin liquid layer, comprising:
treating the treatment surface with a nucleation-phase reactant liquid containing foreign-metal atoms under nucleation-phase reaction conditions; and
thereafter treating the treatment surface by forming a thin liquid layer of a growth-phase reactant liquid containing foreign-metal atoms under growth-phase reaction conditions, the nucleation-phase liquid chemical reaction conditions being different from the growth-phase liquid chemical reaction conditions.
81. A method as in
82. A method as in
83. A method as in
84. A method as in
85. A method as in
86. A method as in
87. A method of depositing a foreign-metal layer onto a base-metal material on a treatment surface of an integrated circuit substrate, comprising:
treating the treatment surface with a nucleation-phase reactant liquid containing foreign-metal atoms under nucleation-phase reaction conditions; and
thereafter treating the treatment surface with a growth-phase reactant liquid containing foreign-metal atoms under growth-phase reaction conditions, the nucleation-phase liquid chemical reaction conditions being different from the growth-phase liquid chemical reaction conditions.
88. A method as in
89. A method as in
90. A method as in
 This application claims the benefit of U.S. Provisional Application Serial No. 60/392,203, filed Jun. 28, 2002.
 The invention is related to the field of integrated circuit fabrication, in particular to methods and apparatuses for the deposition, removal, and treatment of thin films using liquid chemical reactions.
 Electroless plating (or electroless deposition) of copper and other metals has received increasing interest in recent years. This interest is due in part because of the relatively low cost of electroless processes compared to other (e.g., vacuum) deposition techniques, and because of generally surface-controlled, selective, conformal deposition properties of electroless processes. Electroless deposition has a number of potential applications, such as repair of marginal seed layers for copper damascene electroplating, creation of seed layers and barrier layers directly on dielectrics that can be plated, and selective deposition of barrier and electromigration capping layers onto damascene metal (e.g., cobalt and cobalt alloys on copper).
 Conventional electroless metal deposition is conducted in a system containing one or multiple open baths containing plating solution. In a typical operation, a wafer holder immerses a substrate wafer face down in the plating solution during plating operations. The plating solution is exposed to ambient air, especially when the substrate wafer is being moved and the wafer holder does not cover the plating bath surface. Thus, an open bath system has disadvantages. For example, during the metal deposition step, ambient oxygen is readily dissolved in the solution, and the dissolved oxygen can interfere with the desired metal deposition (e.g., by slowing or preventing metal deposition). Electroless plating operations are typically performed at elevated temperatures in a range of 40° C. to 90° C., typically in a range of about 50° C. to 80° C. The plating solution components have a tendency to evaporate. The tendency of water and volatile components to evaporate is exacerbated by the need to ventilate the gaseous spaces over a plating bath, especially to remove explosive or toxic fumes inherent to the electroless solution (e.g., ammonia gas) or created by spontaneous decomposition of its components (e.g., dimethylamine, hydrogen). The heating load caused by evaporation substantially increases the size and costs of a heater required to maintain plating bath temperature. Condensation of evaporate bath constituents on plating-cell walls and on the wafer holder are a source of backside contamination. Maintaining bath concentration, therefore, requires complicated and expensive monitoring and control techniques. See, for example, U.S. Pat. No. 6,537,416, issued Mar. 25, 2003 to Mayer et al., and U.S. patent application Ser. No. 10/272,693, filed Oct. 15, 2002, which are hereby incorporated by reference. A conventional electroless plating bath typically can have a bath volume of 20 liters or more. Typical bath turnover rates required to avoid plate-out and composition drift are 6 hours to 10 hours. Assuming a processing rate of 20 wafers per hour, approximately 160 wafers can be processed with 20 liters.
 A problem of both face-down and face-up plating configurations is hydrogen-bubble entrapment on the plating surface and resulting defects. Hydrogen gas is created as a byproduct of almost all known electroless plating-solution reducing agents. A byproduct of most electroless plating oxidation half-reactions (i.e., the oxidation of the reducing agent) and of the self-degradation of the reducing agents is dissolved molecular hydrogen (H2). As these reactions continue (i.e., plating reactions and bath-aging), the amount of hydrogen increases until the solution becomes saturated and eventually supersaturated with dissolved hydrogen. When this occurs, the formation of hydrogen gas (bubbles) is spontaneous, and occurs most readily on solid interfaces (e.g., vessel walls, wafer surfaces). Areas in which bubbles are attached to the wafer are not plated, creating defects. Therefore, it is advantageous to utilize designs that minimize the propensity for hydrogen formation, or minimize the effective bath age.
 Solution pH influences the reaction rate of the electroless plating process. It is often useful to utilize an alkaline pH-adjuster, for example, lithium-, sodium-, or potassium-hydroxide, but preferably ammonium- or tetramethylammonium hydroxide (“TMAH”) to maintain or adjust the pH. Alkali metal pH-adjusters are inexpensive, but are often unsuitable for semiconductor applications because of their rapid diffusion into and poisoning of various device materials. Ammonium hydroxide is also inexpensive and does not generally degrade device performance, but it is volatile. Therefore, the maintenance of ammonium hydroxide concentration in a plating bath is problematic. TMAH and other analogous organic cation hydroxides do not suff er from either of these problems, but are significantly more expensive. The constituents of a semiconductor electroless plating solution, particularly the reducing agents and TMAH, can be expensive, leading to bath costs in a range of $25/liter to $100/liter. Therefore, one would like to use lower cost materials without the negative impacts. Also, the waste treatment of electroless plating solutions is complicated and expensive. A waste treatment process generally involves forced decomposition of the reducing agents, accompanied by hydrogen gas stripping and dilution. A small amount of dissolved reducing agent can spontaneously breakdown to create a large volume of hydrogen gas in a storage container (an explosive hazard), so the stripping of reducing agents must be driven to completion. A plating solution must also be stripped of metal. The cost of such plating solution post-processing (including capital equipment costs) is typically in a range of $5/liter to $10/liter. Inefficient use of the plating solution, therefore, increases the cost of plating operations significantly.
 Electroless plating solutions are also often inherently unstable. The instability manifests itself in auto-degradation of bath constituents and in the “plating-out” of bath metal as fine metallic particulate in the bulk solution and onto processing equipment walls, filters, and other system components. The presence of plate-out particles also increases the number of defects in the workpieces and diminishes process yield. Generally, the instability of plating solutions increases with reducing agent concentration and with temperature, and decreases with the addition of bath “stabilizers” (e.g., oxygen, chlorine, lead, tin, cadmium, selenium, tellurium). In opposition to this trend, the initiation of electroless plating of a particular metal onto a substrate and the plating deposition rate are also proportional to reducing agent concentration and temperature, and decrease with the addition of bath stabilizers. Thus, plating-solution instability and electroless plating rate and nucleation are inherently linked in a non-advantageous manner.
 Spray techniques have been suggested for electroless plating. See, for example, U.S. Pat. No. 6,065,424, issued May 23, 2000 to Shacham-Diamand et al. In such techniques, reacting plating solution is applied to a wafer surface as a spray or mist. Typically, the wafer is rotating under the spray or mist, and liquid solution is spun radially outwards. Under such conditions, it is difficult to maintain a sufficiently high and uniform reaction temperature because of the simultaneous cooling of the hot fluid by evaporation of the solvent (e.g., water). Alternatively, heating the backside of the wafer by a heated chuck is possible. Nevertheless, this requires a relatively massive element with sufficient heat capacity to maintain a globally uniform temperature over a standard 200 millimeter (mm) or 300 mm wafer. Also, the face-up base of the heating element/chuck is susceptible to chemical contamination and transfer of that contamination to the wafer backside. Furthermore, backside heating does not solve the problem of non-uniform evaporation and cooling of the bath solvent. On the other hand, a wafer chuck should be capable of spinning at high-revolutions per minute (rpm) to enable spin-drying. Splashing of liquid against apparatus walls and misting back onto the product surface can cause contamination of the apparatus and defects on the workpiece. Evaporation and misting of plating solution into the plating space results in substantial loss of the plating solution, and unwanted formation of volatile hazardous chemicals in the effluent.
 Wet processing of isolated conductive-metal circuits connected to transistor elements in the presence of light often encounters a number of processing challenges. One problem is the creation of a photo-induced power source when p-n junctions in the base-circuit transistors are exposed to light. Another problem is the completion of a corrosion circuit on the surface being processed between the exposed isolated metal lines and a processing electrolytic solution. The energy of the light photons is converted to electrical energy, creating a reverse bias potential and a corrosion circuit.
 Thus, liquid chemical reaction techniques, for example, immersion bath and spraying techniques, typically encounter problems such as: difficult or unsuitable control of reaction and process conditions; inability to vary rapidly or dynamically various operating conditions; inability to handle unstable reaction mixtures; accumulation of reaction byproducts; inefficient use of expensive liquid solutions; frequent wafer-handling between process steps; high capital cost of equipment for multi-step processes; and excessive use of valuable clean-room floor space.
 The invention helps to solve some of the problems mentioned above by providing systems and methods for liquid treatment of integrated circuit substrates using a thin liquid layer.
 A novel thin-liquid-layer processing module enables processing of integrated circuit wafers with high throughput and low cost of ownership. Embodiments of such a module are useful for, among others: electroless plating (e.g., deposition of seed layers or the modification of vacuum-deposited seed layers by electroless copper deposition); selective electroless deposition of cobalt and nickel (including combinations of Co, Ni, B, P, and W using electroless process solutions); metal etching (e.g., etching of copper, Ta, TiSN, Co, Ni, etc.); electroless (chemical) polishing (e.g., of copper); various surface treatments (e.g., copper surface reaction with benzotriazole or 3-mercapto-1-propane sulfonic acid); and cleaning and rinsing operations. In particular embodiments in accordance with the invention, a cobalt alloy is electrolessly plated onto copper material in an integrated circuit substrate. An example is a cobalt-capping layer for capping copper.
 The invention is described primarily with respect to its application to electroless plating, but the invention also includes embodiments useful for other liquid treatments, particularly chemical liquid reaction processes and related pretreatment and post-treatment operations. For example, removal of metal layers is also conducted in accordance with the current invention.
 Embodiments in accordance with the invention enable efficient use of small volumes of often unstable fluid reactants and other processing chemicals at elevated temperatures, with preferred embodiments having the ability to recycle these chemicals to reduce operating costs further. Embodiments in accordance with the invention also provide efficient use of surface-cleaning and particle-removing chemicals and the use of minimal water for rinsing operations. Electroless (or chemical) plating, polishing, etching, and rinsing operations are conducted in accordance with the invention with a high degree of global uniformity, using a minimal amount of fluid reactant.
 A thin liquid layer in accordance with the invention is a micro-sized reactor or treatment bath having a volume of the thin fluid gap between a wafer substrate and a treating head. In this specification, therefore, a thin liquid layer for performing a liquid treatment of a substrate surface is sometimes referred to as a “microcell”. The terms “microcell”, “microcell technology”, “microcell module”, and related terms are also used to refer to an apparatus or method in accordance with the invention comprising a treating head that defines a thin fluid gap with a substrate surface, which fluid gap is 10 filled with liquid to form a thin liquid layer. A “supercell” and related terms generally mean a module or apparatus comprising microcell technology combined with the capability of conducting a plurality of pretreatment, cleaning, treatment, and post-treatment operations in a single module, usually without moving a wafer substrate from one station to another.
 In one aspect of the invention, the small volume of a thin liquid layer provides control of the degree or extent of the particular treatment operation. For example, by inserting an aliquot of liquid reactant at a certain concentration into a fluid gap and allowing it to remain in the fluid gap for a time sufficient for a known reaction to run to completion or to an equilibrium point, a controlled known amount of material is deposited on the substrate. For example, a layer having a thickness of 50 nanometers (nm) is deposited by including a known number of moles of reactants in the thin liquid layer sufficient to deposit 50 nm of material, and no more. Similarly, in an etching operation, a desired thickness of material is removed from a substrate surface by including a known number of moles of reactants in the thin liquid layer and allowing them to react to completion.
 In another aspect, such measured deposition, etching, or other treatment operations are conducted in a series of steps. For example, a partial etching is conducted, the substrate's treatment surface is examined, and then a further operation is conducted to complete the etching. In another aspect, a treatment is conducted in a series of steps because a single step operation is undesirable or impossible because of the production of reaction byproducts or for other reasons. For example, in the electroless plating of cobalt on copper, oxidation of the reducing agent generates hydrogen gas. In some embodiments, since the liquid in the thin liquid layer has a limited solubility of hydrogen gas, the liquid is flushed from the fluid gap and replaced with fresh reactants.
 Another advantage of an apparatus and a method in accordance with the invention is that the composition and flowrate of a treatment liquid into a fluid gap is controllable and dynamically variable during treatment operations. In one aspect, certain processes of a substrate treatment, such as nucleation, are conducted under quiescent conditions by injecting an aliquot of reactant liquid into a fluid gap and allowing it to sit. In contrast, certain other processes, such as in a growth phase of electroless cobalt plating, liquid reactant is continuously flowed into the fluid gap, generating convection in the thin liquid layer.
 A microcell is suitable for solving various problems related to electroless plating. In electroless plating techniques, some chemical reactant solutions are chemically unstable. In conventional plating technology, which usually relies on a bath, multiple liters of reactant liquids and other processing liquids are used. When they are unstable and they turn bad, they can no longer be used. In a microcell in accordance with the invention, very small amounts of liquid are used per wafer substrate treated. A conventional immersion bath typically holds a volume of 15 liters to 20 liters. In contrast, the volume of a thin liquid layer in accordance with the invention is in a range of about from 10 milliliters (ml) to 2000 ml, typically 25 ml to 500 ml, and usually 25 ml to 300 ml, depending on wafer size.
 Electroless plating involves a chemical oxidation redox reaction of dissolved metal ions in solution to achieve the desired metal deposition on a substrate. The chemical reaction is typically sensitive to temperature and to pH. A treating head positioned proximate to the substrate wafer forms a fluid gap having a small volume. The fluid gap is filled with liquid reactants or other liquid, depending on the phase of the process. The small volume of the resulting thin liquid layer allows temperature and pH, as well as other process variables, to be controlled and varied effectively. Among other functions, the treating head serves as a pre-heated “thermal mass”, or “heat capacitor”, that heats or cools the reactant fluid and maintains it at a desired temperature. By changing the temperature of a treating head, the temperature of the thin liquid layer is changed to a new temperature.
 Embodiments in accordance with the invention also enable electroless plating in a dark, light-free environment.
 In another aspect, pretreatment, liquid chemical treatment, and post-treatment operations are conducted in the same module, or “supercell”. In another related aspect, a supercell in accordance with the invention comprises a plurality of treating heads for performing multiple operations in a single microcell module.
 In one aspect, a single tube or a plurality of tubes function as liquid inlet tubes into the fluid gap. Typically, the inlet tubes define holes located about the central axis of a treating head so that fluid is injected proximate to the center of the fluid gap and of the treatment surface. Alternatively, the inlet hole or holes are located near a peripheral edge of a treating head creating a type of flow front that moves across a treatment surface of a substrate from one side to the other. This alternative is useful in avoiding the formation of a trapped air pocket or bubbles at the center of a thin liquid layer. In another aspect of the invention, a centrally located showerhead arrangement distributes liquid flow into a fluid gap so that flow is less concentrated at any particular point and so fluid convection is more uniform across a treatment surface. It is found that a showerhead also helps prevent the formation of a trapped bubble during filling.
 In one aspect, a microcell comprises a manifold and a fluid cavity integral with a treating head. In another aspect, to provide balanced distribution of liquid flowthrough the several inlet tubes, a thin piece of diffusion membrane material is placed above the inlet tubes. Flow across the diffusion membrane into the inlet tubes occurs only when there is sufficient pressure differential between the upper and lower side of the diffusion membrane. If there is insufficient pressure differential for flow, no flow occurs. When the pressure differential is achieved, flow occurs relatively evenly across the membrane and into the plurality of inlet tubes and generates a correspondingly balanced liquid flow pattern into and through the fluid gap. In still another aspect, a recirculation tube is in fluidic communication with a liquid source tube from a liquid source and with a liquid inlet tube that leads to a fluid gap. Preferably, a recirculation tube, a liquid source tube, and a liquid inlet tube are connected through a multi-way valve.
 In still another aspect, a manifold bypass tube leads from a manifold cavity and is in fluid communication either with a liquid source, or a drain, or both. A manifold bypass tube allows liquid from a manifold to be recirculated. A manifold bypass tube also functions to release pressure from the manifold cavity, or prevent pressure from forming in the manifold cavity. In another aspect, a manifold cavity typically includes a bubble removal tube for removing gas from the manifold cavity and for releasing pressure. In another aspect, liquid is filtered just priorto entering the fluid gap, usually with a 0.05 micron or 0.1 micron filter (FIG. 9).
 Cobalt and some other metals are ferromagnetic. In one aspect of the invention, magnetic force is used to attract magnetic particles of cobalt (or other metal) and thereby remove cobalt-containing particulate matter from a chemical reactant liquid, from a liquid layer, or from the surfaces of a microcell apparatus. In another aspect, a magnetic field is formed in a microcell to control and focus deposition of cobalt (or other metal) onto a treatment surface. Thus, an electromagnet in the treating head or the substrate holder is used to enhance nucleation, growth, and selectivity. In still another aspect, the magnetic field created by cobalt deposited on a treatment surface (or other magnetic material on a substrate) is measured to determine the amount of material deposited, the thickness of the layer, thickness uniformity, and topography. This allows efficient endpoint determination. In another aspect, continuous measurement of magnetic fields created by deposited cobalt or other magnetic material enables real-time feedback and quality control.
 In still another aspect, light is shown into the fluid gap and an optical sensor measures reflectivity, spectra, or other optical property to measure layer thickness, layer uniformity, and topography.
 In another aspect, a treating head comprises a peripheral edge corresponding substantially in shape to an outer edge of an integrated circuit wafer, and the peripheral edge forms a peripheral slit with the outer edge of the integrated circuit wafer when the wafer is in the substrate holder. The peripheral slit typically comprises a width in a range of about from 0.0 mm to 0.5 mm. A thin fluid gap in accordance with the invention typically comprises a width substantially in a range of about from 0.1 mm to 4 mm. Accordingly, a thin fluid gap typically comprises a volume in a range of about from 30 microliters per cm2 to 300 microliters per cm2 of substrate treatment surface.
 Other features, characteristics and advantages of embodiments in accordance with the invention will become apparent in the detailed description below.
 A more complete understanding of the invention may be obtained by reference to the drawings, in which:
FIG. 1 depicts a microcell device in accordance with the invention in which a treating head in a lowered, operating position and a substrate wafer form a fluid gap filled by a thin liquid layer;
FIG. 2 depicts a microcell device in accordance with the invention in which the treating head is in a raised position;
 FIGS. 3A-3D schematically depict the stages of liquid flowing through an inlet tube to form a thin liquid layer in a fluid gap in accordance with the invention;
FIG. 4 schematically depicts a treating head comprising a megasonic cleaner for removing undesired metal particles from a treatment surface;
 FIGS. 5A-5C schematically depict a rotation of a plurality of treating heads for performing a plurality of surface treatments in the same microcell module in accordance with the invention;
 FIGS. 6A-6C schematically depict the lowering of a megasonic treating head into an operating position in accordance with the invention;
 FIGS. 7A-7E schematically depict a cross-sectional view of treating heads in which the head surface shape is selected to influence temperature or fluid-flow distribution in a thin liquid layer;
 FIGS. 8A-8E schematically depict various designs of inlet-hole locations in the head surface of treating heads in accordance with the invention;
FIG. 9 schematically depicts a microcell module in accordance with the invention comprising a treating head with a manifold cavity and a plurality of inlet tubes;
FIG. 10 schematically depicts a microcell module in accordance with the invention in which liquid reactant is heated, and then flows into a fluid gap, or alternatively is cooled and recirculated to the liquid source;
FIG. 11 schematically depicts a microcell module in accordance with the invention in which liquid flows into a manifold cavity, and then flows through inlet tubes into a fluid gap, or alternatively recirculates to a liquid source, or both;
FIG. 12 schematically depicts an exhaust, or pressure differential, chuck in accordance with the invention for holding a substrate by means of a pressure differential;
FIG. 13 contains a cross-sectional view of a treating head system in accordance with the invention;
FIG. 14 contains a process flow diagram for a microcell apparatus suitable for unstable reaction mixtures; and
FIG. 15 shows the results of electromigration (EM) tests comparing EM lifetime in wafers having Co-capped Cu lines with EM lifetime in baseline wafers having Cu lines with no Co-capping.
 The invention is described herein with reference to FIGS. 1-15. It should be understood that the structures and systems depicted in schematic form in FIGS. 1-14 serve explanatory purposes and are not precise depictions of actual structures and systems in accordance with the invention. For example, the depiction of fluid inlet and outlet streams in the figures below is different from hardware in actual embodiments. Furthermore, the embodiments described herein are exemplary and are not intended to limit the scope of the invention, which is defined in the claims below.
 The terms “liquid treatment”, “treatment”, and related terms are used in a broad sense in this specification to designate any liquid-phase treatment of an integrated circuit substrate, including, for example, pre-treatment operations, cleaning techniques, liquid chemical reactions, rinsing, drying, and post-treatment operations. The term “liquid chemical reaction treatment” is also used in a narrower sense and refers to a treatment conducted at the treatment surface of an integrated circuit substrate involving chemical reaction; for example, deposition, etching, and polishing operations. Broad categories of chemical liquid reaction treatments include electroless metal plating, electroless etching, electrolytic plating, electrolytic etching, metal-oxide deposition, and liquid dielectric deposition.
 The term “dynamically variable” and related terms means that a variable or parameter of an apparatus, method, or composition is variable during a treatment process.
FIG. 1 depicts a planar cross-sectional view 100 of a microcell apparatus 102 in accordance with the invention for conducting a liquid treatment, particularly a chemical liquid reaction treatment, using a thin liquid layer. Microcell apparatus 102 comprises a treating head 104, shown in FIG. 1 in a lowered, operating position. In a lowered, operating position, head surface 106 of treating head 104 forms a fluid gap 108 with a face-up substrate wafer 110. Thus, fluid gap 108 is located between head surface 106 and treatment surface 112 of substrate wafer 110. During liquid treatment of treatment surface 112 in accordance with the invention, fluid gap 108 is filled by a thin liquid layer.
 In this specification, terms of orientation, such as “face-up”, “above”, “below”, “up”, “down”, “top”, “bottom”, and “vertical” used to describe embodiments relate to the relative directions in FIGS. 1-4, 7, and 9-14 in which a substrate wafer defines a substantially horizontal plane. It is understood, however, that the spatial orientation of substrates and apparatuses in embodiments in accordance with the invention are not confined to those depicted in the drawings.
 In a typical electroless process, overtime the plating metal tends to plate onto any available metal surface. The ease of initiation of the plating depends on a number of variables, including roughness, surface oxides, metal catalytic reactivity with the reducing agent, and metal ion reduction charge transfer resistances. Therefore, the presence of metal surfaces of a treating head is problematic. Nevertheless, it is generally desirable to use metal because metals process high thermal conductivity. Therefore, to avoid undesired plating of plating metal onto the metal surface of a treating head, in certain embodiments the exposed head surface is covered with a plastic film, typically having a thickness of about 1 mm or less to minimize interference with heat exchange between the treating head and the thin liquid layer.
 Reactor head 104 comprises a significant mass of a highly conducting material with a heat capacity substantially greater than that of the substrate. Generally, the total (not specific) heat capacity of the head is designed to be more than 10 times greater than that of the substrate, and the thermal conductivity of the heating mass in the head is designed to be as large as possible, generally greater than 0.2 Watt cm−1 K−1. Examples of suitable head materials are metals such as copper (Cu), aluminum (Al), titanium, and iron, particularly aluminum and copper. Because the material for the heating mass of the head may not be compatible with the reacting fluids (e.g., some electroless solutions have a tendency to plate onto a head metal), the bottom surface of the head is typically covered or coated with a thin film of a compatible material (not shown), such as a polyvinylidene difluoride (PVDF), polyethylene (PE), polypropylene (PP), or polytetrafluorethylene (PTFE) coating. The film is sufficiently thick (about 1 mm) to be continuous and to protect the head from spurious reaction and also to resist breaking under handling and typical operation. However, the thin protective film is also sufficiently thin so that it does not substantially reduce the heat-transferring ability of the head to treatment surface 112 (via the thin liquid layer of reactants in the fluid gap between the head and the substrate wafer). Preferably, the flow path of injected liquid through the head is thermally insulated to avoid premature thermal decomposition. A liquid inlet tube 114 made of a suitable material (e.g., plastic) carries liquid to head surface 106 of the head and into fluid gap 108 between head surface 106 and treatment surface 112 of wafer 110. In a preferred embodiment, the liquid is directed to a number of fluid inlet holes (e.g., a “shower head” or distribution outlet) whose location and density are selected to improve and optimize the uniformity of the chemical reaction treatment.
 Microcell apparatus 102 includes a reactor containment vessel 120 having an outer wall 122 and a containment vessel bottom 124. Containment vessel 120 is made from material suitable for withstanding the temperatures and corrosive conditions of plating and etching operations. Examples of suitable materials include polyvinyl chloride (PVC), PVDF, PTFE and various copolymers, PE and PP. Microcell reactor 102 comprises a wafer chuck 130 for supporting substrate wafer 110 having substrate treatment surface 112, which is chemically treated in accordance with the invention. Wafer chuck 130 includes a rotary shaft 132 connected to a motor (not shown) located below containment vessel bottom 124. Wafer chuck 130 also includes three or more support pins 134 for holding wafer 110 above chuck arms 136. Alignment pins 138 are useful for centering wafer 110 during its insertion into microcell module 102 via a wafer-handling robot arm, and are typically tapered to facilitate this operation. Alignment pins 138 also serve to contain wafer 110 from spinning out during operations occurring with rotation (pre-wetting, thin film plating or etching, rinsing, and high speed drying). Designs and uses of a chuck have been described with reference to copper edge bevel removal (EBR) operations, for example, in U.S. Pat. No. 6,537,416, issued Mar. 25, 2003 to Mayer et al., which is incorporated by reference.
 An exit drain 140 is located at bottom 124 of containment vessel 120. Rinse waste and reactant chemical material not otherwise diverted for recycling exit containment vessel 120 through drain 140. The inside bottom 142 of containment vesse 120 is preferably sloped towards exit drain 140 to facilitate draining. In a preferred embodiment, microcell reactor 102 contains a reactant recycling diversion system 150. Diversion system 150 comprises one or more troughs 152 located radially outward from wafer 110. Troughs 152 preferably are movable up or down for alignment substantially with the horizontal plane of wafer 110 to collect fluid 154 emanating from wafer treatment surface 112 in a radial direction (which is induced in large measure by the rotation of the wafer surface). Connected to each of these troughs is a separate drain hole 156. A trough 152 is preferably designed such that fluid is directed downward and into a drain hole 156, typically by locating drain hole 156 at the lowest location of the trough. Drain holes 156 lead to a primary drain tube 158. In a preferred embodiment, primary drain tube 158 fits inside a secondary drain tube 159 with a slightly larger diameter. Thus, a primary drain tube 158 is movable up or down in secondary drain tube 159 and is designed with enough travel to stay always inside secondary tube 159 over its normal length of operational travel. In certain embodiments, a secondary tube 159 is in fluid communication with a treatment liquid source (not shown) for recycling collected treatment liquid to the source. In one aspect, one or more collection troughs are activated for movement up or down, allowing a trough to be positioned relative to a substrate to capture liquid exiting from the thin liquid layer. Alternatively, the fluid gap may be moved with respect to a collection trough by moving a substrate together with the treating head up or down. Preferably, drains associated with each collection trough are separate, thereby allowing liquid collected in one drain to be processed differently from a liquid collected in another drain. For example, cleaning or rinsing liquid is typically discarded, whereas expensive plating liquid is collected and recirculated in certain embodiments. Electroless plating solutions may require expensive waste treatments, so separating concentrated plating waste from more dilute rinses is desirable. The path or trajectory of liquid exiting from a thin fluid gap depends strongly on the rotational speed of the substrate and is used to direct fluid into various troughs. If a substrate is spinning very quickly, liquid exits nearly horizontally. If spinning slowly or not at all (0 rpm to 10 rpm) during some period of the process, then the liquid tends to drip off the side of the wafer and it is collected in appropriately positioned collection troughs or at the bottom of a containment chamber in an exit drain.
FIG. 2 depicts a planar cross-sectional view 160 of microcell apparatus 102 in which treating head 104 is in a raised position. Microcell apparatus 102 optionally includes one or several rinse nozzles 161 or similar applicators for dispensing a thin film of deionized water (DI) or other solution onto substrate surfaces. Preferably, nozzle 161 is located at the periphery of containment vessel 120 to spray liquid inward onto treatment surface 112. For example, DI water or a solution containing a surfactant is useful for pre-wetting a wafer 110, for removing air and entrapped bubbles from the wafer, and for rinsing chemical left after deposition or material-removal operations from a wafer treatment surface 112. Optionally, heated DI is used to improve the efficiency of a liquid reaction treatment by minimizing heating times when hot reactant fluids are used subsequent to DI treatment. An optional rinse nozzle (not shown) directed at backside 162 of a wafer 110 is useful for removing incidental presence of processing fluids from backside 162. Heating a wafer 110 from both the top and bottom with hot DI is useful for preparing a wafer 110 for heated processing, in some cases improving throughput. In some embodiments, after pre-wetting liquid is applied to a wafer 110, a substantial fraction of the total pre-wetting liquid (which may contain DI water, surfactants, dilute acid, or reducing agents), is removed by spinning and is routed to exit drain 140.
FIG. 2 depicts a magnetic source in the form of a magnetic induction coil 180 embedded in treating head 104. In one aspect of the invention, magnet 180 generates a magnetic force to attract magnetic particles of cobalt (or other metal) and thereby remove metal-containing particulate matter from a chemical reactant liquid, from a liquid layer, or from the surfaces of a microcell apparatus. In another aspect, magnet 180 generates a magnetic field to control and focus deposition of cobalt (or other metal) onto a treatment surface. Thus, an electromagnet in the treating head or the substrate holder is used to enhance nucleation, growth, and selectivity. In still another aspect, magnet 180 generates a magnetic field, and a magnetic field sensor 182 proximate to substrate 110 measures the amount of ferromagnetic material deposited, the thickness of the layer, thickness uniformity, and topography. This allows efficient endpoint determination. In another aspect, continuous measurement of magnetic fields created by deposited cobalt or other magnetic material enables realtime feedback and quality control.
FIG. 2 depicts a light source 184 and optical sensor 186. Light source 184 directs light at a substrate 110, and optical sensor 186 measures reflectivity, transmittance, a spectrum, or other optical property to determine layer thickness, layer uniformity, and topography.
 FIGS. 3A-3D schematically depict stages of liquid flowing through an inlet tube to form a thin liquid layer in a fluid gap in accordance with the invention. Treating head 204 with head surface 206 is moved downwards (FIG. 3A) and forms fluid gap 208 between head surface 206 and substrate wafer 210 (FIG. 3B). Reactant liquid 211 flows through liquid inlet tube 214, and makes contact with wafer treatment surface 212 at wafer center 213 (FIG. 3B). The time at which the fluid is turned on, the drop rate of the head to the wafer, and the relative velocity between the head and the rotating (or stationary) wafer are controlled to develop a uniform wetting front emanating from wafer center 213 and growing radially outwards (FIG. 3C). Eventually, gap 208 is filled with liquid, which forms thin liquid layer 220, and drops 222 emanate from peripheral slit opening 224 at a rate equal to the reactant feed rate (FIG. 3D). A particular wetting and filling operation depends on fluid temperature, viscosity, and surface tension, and on the approach velocity and relative rotation rates of treating head 204 and substrate wafer 210.
 The term “flowing liquid into a thin fluid gap” and related terms in this specification are used broadly to refer to several different types of liquid flowing operations. In one sense, flowing liquid into a thin fluid gap means simply flooding the gap or filling it with the liquid to form a thin liquid layer in accordance with the invention. Then, after a thin liquid layer has been formed, the flow of liquid into the thin fluid gap ceases or continues at a same flow rate or continues at a different flow rate. In a second sense, therefore, flowing liquid into a thin fluid gap means continuously flowing liquid, either at steady-state or at an unsteady state, into a thin fluid gap and out of the gap at a corresponding flow rate. It is a feature of some embodiments in accordance with the invention that a liquid treatment can be conducted by filling a thin fluid gap with liquid to form a thin liquid layer, and then cease flow for a period of time, thereby conducting essentially a batch operation. On the other hand, continuous flow operations are conducted in some embodiments.
 The terms “upstream” and “downstream” are used in this specification in their usual sense with reference to directions of process flow streams and relative locations in a process.
 In another aspect, a treating head is heated and maintained at an elevated temperature by one or a plurality of means. In another aspect, an electrical heating element 170 is attached to the top of (FIGS. 1, 2), or embedded into, a treating head. The temperature is controlled by a regulator that senses the head's temperature via thermocouple, thermistor, or similar device embedded in the bulk of the head. Alternatively, a heat exchange manifold with a high-surface-area fluid path interfaces with flow of an externally temperature-controlled head-exchange fluid.
 A treating head rotates with the wafer, opposite to the wafer, or is stationary. Rotation enables modification and control of the hydrodynamics and mass transfer of reactants in the thin liquid layer to the treatment surface. Fluid gap 108, 208 between the substrate and the head is maintained by phenomena similar to those occurring in the fluid bearings. In such an embodiment, the size of the fluid gap is self-regulating. The fluid gap size depends on the relative rotation rate, injected fluid flow rate, and shape of the treating head surface. Alternatively, the fluid gap is controlled by mechanical stops 172 and the like (see FIG. 1). The gap is fine-tuned with a turn-screw against a hard stop, with three position contacts to a support tied to the same base as the chuck mount of the wafer. Thus, the size of the fluid gap and the size of the thin liquid layer are variable.
 In one aspect, pre-wetting and cleaning of the treatment surface is conducted before plating or other chemical liquid reaction treatment. For example, exposing the treatment surface with an activator solution prior to nucleation is conducted using a thin liquid layer in the fluid gap, or alternatively by spraying or otherwise rinsing the treatment surface.
 In another aspect, post-treatment scrubbing of a treatment surface is conducted using megasonic technology. A megasonic transducer is disposed proximate to the substrate treatment surface. In some embodiments, a liquid film is disposed on a substrate treatment surface and then a quartz rod or similar device with a transducer is extended over the treatment surface. The quartz rod vibrates at a very high frequency. The resulting pressure waves in the liquid provide sufficient mechanical energy to clean the treatment surface. For example, a commercially available megasonic device, sold under model name “Goldfinger”, operating at 830 kHz and 125 watts with approximately 50/1000 inch spacing between the unit and the treatment surface provides good cleaning of the substrate. A number of treatments are useful in combination with megasonic agitation in various locations of a microcell reactor for highly efficient cleaning and rinsing of apparatus surfaces. These cleaning treatments typically utilize commercially available proprietary solutions designed to complex with various metal ions, to alter the solution zeta potential, and to modify the surfaces' point of zero surface, thereby freeing the surface of particles.
 Thus, a reactor module in accordance with the invention is useful in combination with a number of other important devices and techniques. FIG. 4 schematically depicts a planar cross-sectional view 300 of an apparatus 302 that contains an alternative treating head 304 comprising a megasonic cleaner 305 for removing undesired metal particles from wafer 310 including treatment surface 312. A vibrating element 316 typically is a rod, plate, or wedge. Vibrating element 316 is connected to a vibration transducer 318, which in turn is connected to a mounting plate or similar aligning fixture 319. The megasonic cleaner is a stand-alone element (as in FIG. 4) or it is integrated into head surface 306 of a treating head. In some embodiments, an apparatus in accordance with the invention comprises a polishing pad and head (not shown). In another aspect, treating head 305 or a polishing pad is moved in close proximity to substrate treatment surface 312 (preferably while it is rotating).
 FIGS. 5A-5C schematically depict a head array 404 comprising a plurality of treating heads for performing a plurality of surface treatments in the same microcell in accordance with the invention. FIGS.5A-5C show the inherent flexibility of a treating head device combined with a face-up rotating wafer module. In FIG. 5A, a microcell reactor head 406 for forming a thin liquid layer in a fluid gap is disposed in a raised position over containment vessel 408. Above the reactor treating head, attached to a rotary main shaft 410 are a microwave cleaning head 412 and treating heads 415, 416. Examples of useful functional types of treating heads have been described above, but embodiments in accordance with the invention are not limited to those particular functions. In one aspect, a plurality of different reactor heads provide different operating conditions to effect different liquid chemical reaction treatments or other liquid treatments. A further example is a treating head that is specifically designed for rinsing the surface; for example, a plastic, unheated head that rotates with the wafer and has outlet holes for cleaning liquid. This is particularly useful when treating a surface that is hydrophobic in nature or has hydrophobic areas. A thin-gap-creating head over a hydrophobic surface makes it possible to maintain complete wetting using very little rinse water to clean surfaces. FIG. 5B shows partial rotation around rotary main shaft 410. FIG. 5C shows completed rotation of cleaning head 412 into a raised position above containment vessel 408. FIGS. 6A-6C schematically depict the lowering of megasonic treating head 412 into an operating position in accordance with the invention. In FIG. 6A, head 412 is in a raised position above containment vessel 404. FIG. 6B shows a partially lowered array 404, and FIG. 6C shows array 404 in a lowered position in which megasonic treating head 412 is disposed near to a treatment surface of a substrate wafer.
 As depicted in FIGS. 1 and 2, treating head 104 is characterized by a head surface 106 having a shape that is substantially flat and horizontal. In other embodiments, a treating head has a head surface that has been shaped to control or improve process conditions; for example, to effect a more uniform temperature distribution or a desired fluid-flow distribution in a thin liquid layer. FIG. 7A schematically depicts a cross-sectional view of a treating head 450 with a head surface 452 that is upwardly conical, typically forming an angle to the horizontal plane of substrate 454 in a range of about from 1 degree to 30 degrees. A head surface shape as in FIG. 7A forms a narrow slit 456 at its peripheral edge 457 with outer edge 458 of substrate wafer 454, while the upward conical surface increases the volume of fluid gap 459 compared to the volume formed by a flat head surface. FIG. 7B schematically depicts a cross-sectional view of a treating head 460 with a central head surface 461 that is upwardly conical and with an outer head surface 462 that is substantially flat and horizontal. A head surface shape as in FIG. 7B forms a narrow fluid gap 464 with substrate 466 at its outer head surface 462, while upward conical surface 461 increases the volume of the fluid gap compared to the volume formed by a flat head surface. FIG. 7C schematically depicts a cross-sectional view of a treating head 470 having a central head surface 472 that is substantially flat and horizontal, and having an outer head surface 473 that is substantially flat and horizontal. When outer head surface 473 is positioned very near a substrate surface, it forms a narrow peripheral slit 474, but the width of fluid gap 475 corresponding to central head surface 472 is wider and provides a larger volume. FIG. 7D schematically depicts a cross-sectional view of a treating head 480 having a head surface 482 that is substantially convex relative to a face-up substrate treatment surface 484. FIG. 7E schematically depicts a cross-sectional view of a treating head 490 having a central head surface 492 that is downwardly conical, and having an outer head surface 494 that is substantially flat and horizontal relative to substrate 496. A treating head 490 forms a fluid gap 497 that is thin at the center of a wafer 496, and is relatively thick at the edges of the wafer.
 FIGS. 8A-8E schematically depicts bottom views of treating heads showing various designs of inlet-hole locations in the head surface of treating heads in accordance with the invention. FIG. 8A depicts treating head 510 having head surface 512 with a centrally located inlet hole 514. FIG. 8B depicts treating head 520 having head surface 522 with a centrally located showerhead arrangement of inlet holes 524 that distributes liquid flow into a fluid gap so that it is less concentrated at any particular point and so fluid convection is more uniform across a treatment surface. FIG. 8C depicts treating head 530 having head surface 532 with a plurality of inlet holes 534 extending from center 536 on a radial path outwards to edge 538. FIG. 8D depicts treating head 540 having head surface 542 with a plurality of inlet holes 544 in a substantially straight line through the center of 546. FIG. 8E depicts treating head 550 having head surface 552 with an inlet hole 554 located near a peripheral edge, creating a type of flow front that moves across a treatment surface of a substrate from one side to the other. This alternative is useful in avoiding the formation of a trapped air pocket or bubbles at the center of a thin liquid layer.
FIG. 9 schematically depicts a cross section 600 of a microcell module 602 in accordance with the invention. Microcell module 602 comprises a microcell-reactor treating head 604 and a chuck 606 for holding a substrate wafer 608. When a substrate wafer 608 is present in substrate holder 606, treating head 604 and substrate 608 define a thin fluid gap 612 that is located between head surface 614 of treating head 604 and treatment surface 616 of substrate 608. Treating head 604 comprises a manifold cavity 618 and a plurality of inlet tubes 620 that provide passage of fluid from manifold cavity 618 through inlet holes 622 into a fluid gap 612. Thus, treating head 604 serves as a showerhead-type inlet manifold for liquid and gaseous fluids into thin fluid gap 612. Microcell 602 further comprises a manifold inlet 624 through which fluid flows into manifold cavity 618. In liquid treatment methods in accordance with the invention, liquid in fluid gap 612 forms a thin liquid layer 626. Typically, the combination of treating head 604 and substrate-holding chuck 606 are contained within a containment chamber 630 when treating head 604 is in a lowered position, as depicted in FIG. 9, proximate to substrate wafer 608 to form thin fluid gap 612. Containment chamber 630 comprises an exit drain 631. In certain preferred embodiments, treating head 604 comprises zoned heaters 632 controlled by conventional means. In another aspect, as depicted in FIG. 9, a substrate-holder chuck 606 comprises backside dispensing tubes 634 for directing heating fluid, deionized water, cleaning liquids or other liquids 635 at backside 636 of a substrate 608. In certain embodiments, chuck 606 includes backside zone heaters 638 for heating substrate 608. A multizoned heater generates and controls a nonuniform heating profile in treating head 604 or in a substrate holder 606, and thereby enhances temperature control in a thin liquid layer 626. Time-varying control of a heater allows dynamic variation of temperatures during the treatment process.
 Microcell apparatus 602 further includes a first source tube 640 and a second source tube 642 from first liquid source 644 and second liquid source 646, respectively. In certain preferred embodiments in accordance with the invention, a plurality of liquid streams are mixed at or near the fluid gap, allowing an unstable chemical mixture to be formed at or near the point of use and thereby avoid premature reaction or decomposition. Accordingly, as depicted in FIG. 9, microcell apparatus 602 includes point of use mixer 648. First, and second liquid streams in first and second source tubes 640, 642, respectively, flow into mixer 648 where they are mixed. The mixed liquid stream exits mixer 648 and flows through manifold inlet 624 into manifold cavity 618.
 Treating head 604 functions as a showerhead-type injection manifold and thereby provides distributed flow of liquid into fluid gap 612. In another aspect, a diffusion membrane 650 located in manifold cavity 618 balances liquid flow from manifold cavity 618 through inlet tubes 620 into fluid 612. Typically, a diffusion membrane 650 rests on manifold cavity bottom 652 and covers inlet tubes 620. In another aspect, a hydrophilic membrane 650 maintains bubble-free wetting of a substrate treatment surface 616. Preferably, a diffusion membrane 650 is selected so that flow across the diffusion membrane into the inlet tubes occurs only when there is sufficient pressure differential between the upper and lower side of the diffusion membrane. If there is insufficient pressure differential for flow, no flow occurs. When the pressure differential is achieved, flow occurs relatively evenly across the membrane and into the plurality of inlet tubes and generates a correspondingly balanced liquid flow pattern into and through the fluid gap.
 In still another aspect, microcell apparatus 602 comprises a manifold bypass tube 660 that leads from manifold cavity 618 and is in fluidic communication either with a liquid source, or a drain, or both. Manifold bypass tube 660 allows liquid from a manifold to be diverted and recirculated. Manifold bypass tube 660 also functions to release pressure from the manifold cavity, or prevent pressure from forming in the manifold cavity. By appropriate control of manifold bypass valve 662, liquid flow through manifold cavity 618 into thin fluid gap 612 is controlled. Although manifold bypass 660 is depicted in FIG. 9 at the side of treating head 604, in practice manifold bypass tube 660 is preferably located at the center 664 of treating head 604 to allow rotation of treating head 604. Chuck 606 is connected to rotary shaft 666, which provides rotation of substrate 608.
FIG. 10 schematically depicts a cross-section 700 of a microcell module 702 in accordance with the invention. Microcell module 702 comprises a microcell-reactor treating head 704 and a chuck 706 for holding a substrate wafer 708. Treating head 704 and chuck 706 are disposed within containment chamber 710, which has an exit drain 711. When a substrate wafer 708 is present in substrate holder 706, treating head 704 and substrate 708 define a thin fluid gap 712 that is located between head surface 714 of treating head 704 and treatment surface 716 of substrate 708.
 A liquid inlet tube 720 passes through treating head 704 and inlet hole 722 into fluid gap 712. During operation, fluid gap 712 is filled with liquid to form a thin liquid layer 715 in accordance with the invention. A liquid source tube 724 is in fluidic communication with liquid inlet tube 720 and recirculation tube 726 through 3-way source valve 728. Liquid reactant is present in liquid source 730. Liquid source tube 724 is in fluidic communication with source exit 732 of liquid source 730. In certain preferred embodiments, microcell module 702 comprises liquid heater 734, which heats liquid from liquid source 730 as it flows into liquid source tube 724. In more preferred embodiments, microcell module 702 comprises a liquid cooler 736 for cooling a heated recirculated liquid flowing in recirculation tube 726 back into liquid source 730. Valve 728 is controlled to control the portion of heated liquid in liquid source tube 724 that flows through liquid inlet tube 720 into thin fluid gap 712. Valve 728 is controllable so that all, or none, or just a portion of liquid flowing in liquid source tube 724 flows into liquid inlet tube 720. Preferably, valve 728 is dynamically controllable so that the amounts of liquid in liquid inlet tube 720 and recirculation tube 726 are dynamically variable during treatment operations. In certain embodiments, microcell apparatus 702 includes a liquid recycle tube 740 with dynamically controllable recycle valve 744 for recycling collected liquid reactant 742 from the bottom of containment chamber 710 (or from a diversion system 150, as described with reference to FIG. 1) back to liquid source 730.
 Reactant liquid source 730 optionally includes a heat exchanger for cooling (to stop autocatalytic reactions) or heating (preparing the chemical for reaction and subsequent recycling). The choice of heating, cooling, recirculating, and recycling operations within microcell 702 in methods in accordance with the invention is determined by the properties of the particular materials/chemicals being used, their stability, and processing temperature. Chemicals and mixtures that are highly unstable are kept cool and then heated only right before application to the wafer. For example, in certain embodiments, cool fluid is pumped from a liquid source 730, heated in-line in heater 734, and/or heated in the intricacies of any feed lines imbedded in a heated treating head, and then introduced into the thin fluid gap beneath treating head. While the fluid is in the gap, it is also optionally being heated by a heated head 704. After use, the hot liquid is ejected from gap 712 and recycled back to a cooled containment vessel via a recycling tube 740 or a reactant recycling diversion apparatus 150, as described with reference to FIG. 1. Alternatively, collected liquid exits through drain 711 for waste disposal. More stable chemistries are maintained in liquid source 730 at or near an operating temperature desired in a thin liquid layer 715. Commonly, a “bleed and feed” of the fluid in the reactant vessel is employed to avoid substantial changes to bath properties due to consumption of reactants in the process, dilution due to rinse water introduction to the bath, and auto-decomposition. Determination of the optimum liquid-source size and turnover rate is made according to reactant solution stability measurements consistent with the liquid-source turnover time (i.e., removal rate (liter/hr)/source volume (liter)). Removal rate calculations include considerations of evaporation, consumption, decomposition, rinse-water introduction, and efficiency of reactant-collection in a recycling diversion apparatus.
 In another aspect, treating head 704 typically comprises a gas injection tube 746 for injecting inert gas into fluid gap 712 to eject thin liquid layer 715. Gas injection tube 746 is also useful for removing gas bubbles from thin liquid layer 715 in fluid gap 712 and for releasing pressure from thin liquid layer 715.
 Preferably, chuck 706 includes multizone heaters 747 for heating substrate wafer 716.
 In another aspect, liquid is filtered just prior to entering the fluid gap, usually with a 0.05 micron or 0.1 micron filter 748 located in liquid inlet tube 720.
 In another aspect, microcell 702 preferably includes a plurality of inert gas nozzles 752 located for creating an inert-gas blanket (e.g., nitrogen) around the outer edge of the wafer to inhibit or prevent air oxidation of the wafer surface and thin liquid layer 715.
 Chuck 706 is connected to rotary shaft 766, which provides rotation of substrate 708.
 A thin liquid layer typically completely covers the treatment surface of a wafer substrate, even very hydrophobic surfaces, due to the proximity of the head surface to the substrate and to the combined properties of liquid viscosity and surface tension. The liquid in a thin fluid gap in accordance with the invention typically does not exit the surface of the wafer unless some external force is applied to it. Under typical process conditions, liquid does not begin to exit from the open edge of a thin liquid layer in a fluid gap having a peripheral-slit width of 1 mm to 2 mm until the substrate wafer is rotated about 50 rpm or greater. In some embodiments, gas is injected into the fluid gap to expel liquid from the fluid gap, thereby removing the surface tension and any associated suction between the treating head and a substrate wafer.
FIG. 11 schematically depicts a cross-section 800 of a microcell module 802 in accordance with the invention. Microcell module 802 comprises a microcell-reactor treating head 804 and a chuck 806 for holding a substrate wafer 808. When a substrate wafer 808 is present in substrate holder 806, treating head 804 and substrate 808 define a thin fluid gap 812 that is located between head surface 814 of treating head 804 and treatment surface 816 of substrate 808. Treating head 804 comprises a manifold cavity 818 and a plurality of inlet tubes 820 that provide passage of fluid from manifold cavity 818 through inlet holes 822 into a fluid gap 812. Thus, treating head 804 serves as a showerhead-type inlet manifold for liquid and gaseous fluids into thin fluid gap 812. Microcell 802 further comprises a manifold inlet 824 through which fluid flows into manifold cavity 818. In chemical liquid reaction treatment or other liquid treatment methods in accordance with the invention, liquid in fluid gap 812 forms a thin liquid layer 826. Typically, the combination of treating head 804 and substrate-holding chuck 806 are contained within a containment chamber 830 when treating head 804 is in a lowered position, as depicted in FIG. 11, proximate to substrate wafer 808 to form thin fluid gap 812. Containment chamber 830 comprises an exit drain 831. In certain preferred embodiments, treating head 804 comprises zoned heaters 832. In another aspect, as depicted in FIG. 11, a substrate-holder chuck 806 comprises backside dispensing tubes 834 for directing heating fluid, deionized water, cleaning liquids, or other liquids 835 at backside 836 of a substrate 808. In certain embodiments, chuck 806 includes backside zone heaters 838 for heating substrate 808. A multizoned heater generates and controls a nonuniform heating profile in treating head 804 or in a substrate-holder 806, and thereby enhances temperature control in a thin liquid layer 826. Time-varying control of a heater allows dynamic variation of temperatures during the treatment process.
 Microcell apparatus 802 further includes a liquid source tube 840 from liquid source 844. Liquid source tube 840 is in fluidic communication through manifold inlet 824 with manifold cavity 818.
 Treating head 804 functions as a showerhead-type injection manifold and thereby provides distributed flow of liquid into fluid gap 812. In another aspect, a diffusion membrane 850 located in manifold cavity 818 balances liquid flow from manifold cavity 818 through inlet tubes 820 into fluid 812. Typically, a diffusion membrane 850 rests on manifold cavity bottom 852 and covers inlet tubes 820. In another aspect, a hydrophilic membrane 850 maintains bubble-free wetting of a substrate treatment surface 816. Preferably, a diffusion membrane 850 is selected so that flow across the diffusion membrane into the inlet tubes occurs only when there is sufficient pressure differential between the upper and lower side of the diffusion membrane. If there is insufficient pressure differential for flow, no flow occurs. When the pressure differential is achieved, flow occurs relatively evenly across the membrane and into the plurality of inlet tubes and generates a correspondingly balanced liquid flow pattern into and through fluid gap 812.
 In still another aspect, microcell apparatus 802 comprises a manifold bypass tube. 860 that leads from manifold cavity 818 and is in fluidic communication through bypass valve 862 with recirculation to 864. Manifold bypass tube 860 allows liquid from a manifold to be diverted and recirculated. Manifold bypass tube 860 also functions to release pressure from the manifold cavity, or prevent pressure from forming in the manifold cavity. By appropriate control of manifold bypass valve 862, liquid flow through manifold cavity 818 into thin fluid gap 812 is controlled. Although manifold bypass 860 is depicted in FIG. 11 at the side of treating head 804, in practice manifold bypass tube 860 is located at the center 864 of treating head 804 to allow rotation of treating head 804. Chuck 806 is connected to rotary shaft 866, which provides rotation of substrate 808.
 In another aspect, apparatus 802 typically comprises a gas injection tube 868 located in treating head 804 for injecting inert gas into fluid gap 812 to eject thin liquid layer 815. Gas injection tube 868 is also useful for removing gas bubbles from thin liquid layer 815 in fluid gap 812 and for releasing pressure from thin liquid layer 815.
 Thus, in one aspect of the invention, a liquid layer in a fluid gap is expelled by injecting gas into the fluid gap using sufficient pressure. Alternatively or additionally, the chemicals in the gap are flushed by flowing water through manifold cavity 818 and inlet tubes 820. This flushing is typically followed with expulsion of the rinse water from the gap by injecting gas into fluid gap 812. When injected proximate to the center of the fluid gap, the gas displaces the liquid radially outwards. The trajectory of the injected liquid depends to a large degree on the pressure and flowrate of gas into the fluid gap and on the rotational speed of the substrate. Using corresponding collection troughs and exit drains as described above, ejected fluid is recycled, processed to recover components, processed to expel undesired byproducts, and recycled or discarded. In this manner, an expensive plating solution is used again. In a rinse phase, the treating head is still down. An objective of the rinse phase is to clean the wafer of residual chemicals, as well as clean the head. The treating head is rinsed to avoid evaporation of plating solution and resultant drying and plating of chemicals on the treating surface of the head. Typically, deionized rinsing water is injected through the treating head into the fluid gap and the substrate is spun at a relatively high rotation rate. Multiple rinses can be beneficial. In another aspect, the rotation rate is thereafter slowed down and the treating head is lifted. Preferably, the treating head is further cleaned in its upper position with a spray stream of deionized water or other cleaning liquid.
 In an exemplary embodiment of the invention, a fluid gap is filled (flooded) to form a thin liquid layer on a 200 mm substrate wafer by spinning the substrate at a speed in a range of about from 10 rpm to 50 rpm, preferably about 25 rpm, and flowing liquid at a flowrate of about from 100 ml per minute to 500 ml per minute into the center of the fluid gap. Alternatively, a thin liquid layer is formed by flowing liquid from either a single hole or manifold with essentially the same rotation rate and flowrate, but flowing the liquid from an edge. During a plating operation, the rotation rate is typically in a range of from 0 rpm to about 100 rpm, and the liquid flowrate is anywhere from 0 ml per minute to 500 ml per minute.
 Certain embodiments in accordance with the invention include a substrate holder that comprises an exhaust chuck, in which a substrate is held in the chuck by means of a pressure differential. FIG. 12 depicts schematically a cross-sectional view 870 of an exhaust chuck system 872 located in containment chamber 874. As depicted with its essential elements in FIG. 12, exhaust chuck system 872 includes substrate supporter 876 that supports a substrate wafer 878. An exhaust tube 880 is located in lower containment chamber 882. Typically, air or other gas is drawn from the bottom of containment chamber 874 through exhaust tube 880. An annular collar 884 in containment chamber 874 is located proximate to substrate 878, usually just below it, restricting gas flow from upper containment chamber 885 above the substrate to lower containment chamber 882 below the substrate. Annular collar 884 forms a narrow flow passage 886 between its inside collar edge 888 and outer wafer edge 890 of substrate wafer 878. An annular seal 892 creates a seal between an outer collar lip 893 and a chamber wall lip 894 when chuck system 872 is in a closed, chucking position, as depicted in FIG. 12. Narrow flow passage 886 typically has a width in a range of about from 1 mm to 7 mm. The constriction of flow causes a pressure differential, the pressure above substrate 878 being greater than the pressure below it. The pressure differential creates a chucking action that holds substrate 878 on substrate support 876. This is an alternative to a conventional vacuum chuck, which pulls on the wafer and requires a rotary union inside to maintain the vacuum. The exhaust flow through exhaust tube 880 is relatively uniform and thereby pulls the substrate uniformly down on supporting chuck base 876. With a 4-mm-flow-passage 886, about 1 meter per second of air flowing through narrow flow passage 886 produces about 20 pounds of force on a 200 mm wafer substrate. In another aspect, annular collar 884 is moved down (or up, depending on the particular configuration) for wafer handling using a collar-moving mechanism on upper chamber wall 895, which includes rotatable wheel 896. An example of a moving mechanism of a chuck is described in U.S. Pat. No. 6,126,382, issued Oct. 3, 2000 to Scales et al., which is hereby incorporated by reference. As annular collar 884 is moved down, narrow flow passage 886 is opened, and the pressure differential is thereby removed. On the other hand, the exhaust flow through exhaust tube 880 is typically maintained. An advantage of an exhaust chuck is that it eliminates clamps, pins, and other structural clamping and aligning structures that cause defects and interfere with maintaining and removing a thin liquid layer (not shown) on substrate 878. Preferably, annular collar 884 includes drain holes 897 for liquid drainage.
FIG. 13 contains a cross-sectional view 900 of a preferred treating head system 902 in accordance with the invention. System 902 includes a treating head 904, a substrate holder 906 for holding a substrate wafer 908, and a dam or fluid-bearing ring 910. Fluid-bearing ring 910 is connected to treating head 904 and forms a peripheral edge of treating head 904. Typically, fluid=bearing ring 910 is an integral part of treating head 904. During operation, with treating head 904 in a lowered position, fluid-bearing ring 910 and treatment surface 911 of wafer 908 define a peripheral slit 912 near outer edge 913 of substrate wafer 908. Fluid-bearing ring 910 typically comprises a radial thickness of about 1 mm. Fluid-bearing ring 910 is typically disposed in a range of about from 0.25 mm to 3 mm radially inwards from outer edge 913 of wafer substrate 908. Fluid-bearing ring 910 typically protrudes about 0.1 mm to 1.5 mm below the plane of head surface 914 of treating head 904. Fluid-bearing ring 910 substantially encloses a fluid gap 915 defined by head surface 914 of treating head 904 and treatment surface 911 of substrate wafer 908. Fluid-bearing ring 910 constricts or prevents the flow of liquid from a thin liquid film out of fluid gap 915. Peripheral slit 912 comprises a width (vertical dimension in FIG. 13) between fluid-bearing ring 910 and treatment surface 911 in a range of about 0.0 mm to 0.3 mm. In some embodiments, when treating head 904 and substrate wafer 908 are stationary or are rotated in the same direction at the same rotation rate, peripheral slit 912 is virtually zero. In other embodiments, when treating head 904 and substrate wafer 908 are not rotated together, peripheral slit 912 has a width greater than zero, even though infinitesimal. Practically, a positive pressure differential between a thin liquid layer present in a filled-fluid gap 915 and the region radially exterior to fluid-bearing ring 910 causes the width of peripheral slit 912 to be greater than zero. In preferred embodiments, as depicted in FIG. 13, treating head 904 comprises an upper region 916 having an increased diameter 917 compared to outside diameter 918 of fluid-bearing ring 910. Increased diameter 917 results in an increase in the thermal mass of treating head 904, which is useful for maintaining a desired temperature of a thin liquid film in fluid gap 915. A fluid-bearing ring 910 is useful for several reasons. A fluid-bearing ring 910 makes control of the width of thin fluid gap 915 easier. Also, the tendency for liquid in fluid gap 915 to drain is reduced, particularly when using a chuck or other apparatus that makes contact with the wafer edge and creates a “channel” for liquid to drain out of fluid gap 915. Specifically, fluid-bearing ring 910 allows fluid to drain out of the region external of the ring, but prevents or inhibits fluid from flowing from fluid gap 915 through peripheral slit 912. Also, a fluid-bearing ring 910 confines fluid in fluid gap 915 under conditions of high wafer and/or head rotation. A high rotation rate enables a significant increase in convection of mass and heat without liquid in fluid gap 915 being ejected from the gap because a peripheral slit between ring 910 at the peripheral edge of treating head 904 and the substrate below is very narrow or even closed. Furthermore, flushing of chemicals and rinse water is greatly improved. In a preferred embodiment, gas is introduced into the center of fluid gap 915 to expel a thin liquid layer or other liquid, and a gas bubble is thereby created. The presence of a fluid-bearing ring 910 allows slow expansion of the bubble and controlled expulsion of liquid from fluid gap 915 through slit 912. In preferred embodiments, wafer substrate 908 and treating head 904 are both spinning at the same rate, preferably in a range of about 300 rpm to 800 rpm, in the same direction. A pressurized gas bubble is introduced into the center of the water and is slowly allowed to force fluid out of slit 912 created by fluid-bearing ring 910. The centripetal forces on the fluid entrained on the wafer and head allow these surfaces to be rapidly drained of excess fluid, and thereby reduces the likelihood of fluid dripping back down from treating head 904 onto wafer 908. In a case in which wafer 908 has both a hydrophobic area (e.g., low-k Coral™ dielectric) and a hydrophilic surface area (e.g., copper metal), an apparatus having a treating head system 902 and a high rpm process for expelling liquid from fluid gap 915 is particularly useful for reducing the incidence of chemical defects and water-spot defects (e.g., streaks, stains, contamination).
 The pH often tends to drift during cobalt plating. For example, in some embodiments, the liquid plating solution contains an ammonium ion and the solution is heated to an elevated temperature. In aqueous solutions, the ammonium ion is in equilibrium with the amount of dissolved ammonia gas by the chemical equilibrium:
NH4 ++OH−⇄NH3(dissolved gas)+H2O.
 Since warm dissolved ammonia gas has a relatively high partial vapor pressure, it tends to evaporate very quickly into any air above the bath. Continual or continuous adjustment of pH and other liquid properties is done with a multiple chemistry flow capability by which chemical constituents of a reactant solution are contained in a plurality of reactant liquid sources, and the flow rate from one or a plurality of liquid sources into the fluid gap is dynamically variable.
 In another aspect, a post-treatment anneal of a substrate wafer is conducted in an apparatus in accordance with the invention, either at low temperature up to about 400° C., or up to about 1000° C. using a very rapid thermal process (RTP) heater.
 In another aspect, the chemical composition of a plating solution is varied during plating operations so that the chemical composition of the deposited metal layer varies spatially within the layer. Such a structure may alternatively be viewed as a multilayer structure. For example, phosphor, tungsten, or boron are known to improve the barrier properties of cobalt. In accordance with certain aspects of the invention, a particular cobalt alloy is deposited at the bottom of a capping layer, a different cobalt alloy is formed for the bulk of the capping layer, and a third cobalt alloy provides the top of a capping layer.
 A process flow diagram for a microcell apparatus 920 suitable for unstable reaction mixtures is depicted in FIG. 14. A liquid reaction mixture is formed by mixing liquid streams from two separate liquid sources. A first liquid source 924 contains components of the reaction mixture that are inherently stable at an elevated temperature. For example, in the case of an electroless plating solution, liquid source 924 contains metal ions, buffers, complexing agents, surfactants, etc. A second liquid source 926 contains unstable or destabilizing components of the reaction mixture that cause the bath to react and potentially to decompose prematurely. For example, in the case of electroless plating, second liquid source contains reducing agents. A containment vessel 928 encloses treating head 922 and wafer chuck 930 supporting a wafer substrate 932. In a closed position, treating head 922 forms a thin fluid gap 934 with substrate 932. In some embodiments, the material in liquid source 924 is heated by a heater 935 to a temperature slightly higher than the desired operating temperature in microcell thin film layer 936 located in fluid gap 934. The higher temperature value in liquid source 924 is controlled such that when liquid from source 924 is mixed with an appropriate amount of cooler liquid (e.g., reducing agents) from liquid source 926, the temperature of the reaction mixture after mixing is substantially the same as the desired temperature in thin liquid layer 936 (which is typically the same as the temperature of treating head 922). In process 920, a first liquid stream is pumped from first liquid source 924 in first source tube 938 by pump 940 through filter 942 for continuous filtration, and then through a flow meter/controller 944 into three-way recirculation valve 946. Recirculation vale 946 is dynamically controllable so that all or a portion of first liquid stream 938 is diverted into recirculation stream in recirculation tube 948. Undiverted first liquid stream 950 continues towards microcell fluid gap 934. Downstream from three-way valve 946, a tee 952 into line 950 allows the introduction of either DI water or pressured gas (e.g., N2) for forcing the removal of fluid from microcell gap 934. Further downstream, a tee 954 allows injection of a second liquid stream in second source tube 956 from second liquid source 926 into mixture line 958. For example, in electroless plating, second liquid stream 956 comprises a relatively small volume of concentrated reducing agents. Methods of introducing pressurized reducing agent through tee 954 into mixture line 958 include the use of a gear or syringe pump 960. By controlling the concentrations in liquid sources 924, 926 and the flow rates in liquid streams 950, 956, the concentrations of bath components in mixture line 958 are accurately controlled. In some cases, it is useful to introduce an inline mixing chamber downstream of tee 954 to properly mix the reaction mixture upstream of microcell inlet tube 962.
 In still another aspect, modules comprising microcell (thin liquid layer) devices in accordance with the invention are stacked substantially vertically. This enables higher throughput of substrates per unit surface area of manufacturing floor space. Since many pre-treating, rinsing, and post-treating operations are performed in a single microcell in accordance with the invention, utilization of production space improves.
 Although embodiments in accordance with the invention are described herein mainly with respect to electroless plating techniques, it is clear that an apparatus and a method in accordance with the invention-are useful for many types of wet substrate treatments. Various substrate treatments include liquid chemical reactions as well as non-reactive treatments (e.g., pretreatment cleaning and rinsing). In an important aspect of the invention, a plurality of substrate treatments of a substrate are conducted sequentially in one module or supercell. An example of a liquid chemical reaction treatment is the uniform etching of a substrate surface. Another example is the stripping of an oxide from a substrate surface. For example, a protective oxide layer or an oxide layer that simply formed in an oxidizing environment is stripped off before electroless plating operations begin. In another aspect, metal particles and other contaminants are cleaned from a treating head, containment chamber walls, and other surfaces prior to liquid chemical reaction treatment.
 Electroless plating is an autocatalytic plating technique, the process physics enabling selective deposition of a metallic coating by a controlled chemical reduction that is catalyzed on metal or alloy being deposited. Electroless plating depends on the action of a chemical reducing agent in solution to reduce metallic ions to the metal. Unlike a homogeneous chemical reduction, the plating reaction takes place only on “catalytic” surfaces rather than throughout the plating solution. The process occurs by the simultaneous reduction of metal and the oxidation of a “reducing agent” on the metal surface. These two processes do not have to occur at exactly the same place on the metal surface, but there must be an electrical path between the location of the reducing agent's oxidation (generating excess electrons) and the location of metal deposition (combining the generated electrons with metal ions). Electroless plating has been used for depositing a large number of materials, including Cu, Ni, Co, Fe, Pd, Pt, Ru, Rh, Au, Ag, Sn, Pb, as well as alloys containing these metals, plus Mn, Mg, W, P and/or B. Various metals are deposited electrolessly onto an electronic component, including, for example, copper, nickel, cobalt, gold, silver, palladium, platinum, rhodium, cobalt, tungsten phosphorous, and combinations thereof. Electroless polymerization is performed by an analogous electroless process for some conductive polymers (e.g., polyanaline). Typical chemical reducing agents include ammonium hypophosphite ((NH4)H2PO2), formaldehyde (CH2O), hydrazine, borohydride, dimethylamine borane-(DMAB), glyoxylic acid, redox-pairs (e.g., Fe(II)/Fe(III)) and derivatives of these. A chemical reducing agent in plating solution is a source of electrons for the reduction of a metal ion to a deposited metal atom on the surface:
M n+ ne=M 0
 where Mn+ represents a reducible metal ion in the solvent (typically water). Complexing agents (e.g., acetate, succinate, malate, malonate, citrate, etc.) are often used in plating solutions to enhance solubility at pH values where the metal ion would otherwise be insoluble. Complexation of the metal is also useful for shifting the potential of deposition to obtain desirable conditions for deposition.
 In some cases, a particularly strong and catalytically-active reducing agent is important at the beginning of an electroless plating process in order to initiate plating of the metal onto the substrate surface. This is even more important in cases where the initiation of the plating process is on a foreign metal surface (e.g., initiation of cobalt electroless deposition onto a copper surface).
 Accordingly, preferred embodiments in accordance with the invention involving electroless plating on a foreign substrate (e.g., cobalt on copper) typically comprise a two-phase process including a nucleation phase and a growth phase. In the nucleation phase, a desired depositing metal (e.g., cobalt) is caused to deposit on a foreign metal substrate surface (e.g., copper). Afterwards, in the growth phase, the desired metal (e.g., cobalt) grows on a film of similar metal (e.g., cobalt). Typically, optimum or idealized process conditions for the nucleation phase are different from those of the growth phase. For example, for electroless plating of cobalt on copper, the optimal set of conditions for the nucleation reaction to occur is very different from that of the growth reaction. Nucleation of cobalt onto a copper substrate involves the generation of excess reduced cobalt-metal atoms at the copper surface at a sufficient concentration for formation of a nucleation layer of cobalt. To create this concentration of surface cobalt atoms, a reducing agent of sufficient strength (i.e., an agent having suitable free-energy driving-force and kinetics) to reduce sufficient metal ions at a sufficiently rapid rate is required. One example of such a reducing agent is mopholine borane. Because the process of cobalt-ion reduction is likely stepwise, the creation of partially-reduced surface-absorbed metal ions presents a problem. The partially-reduced ions can diffuse away into the electrolyte and not aid in the nucleation process. To minimize this possibility, initiation of the electroless plating operation during a nucleation phase is typically performed under stagnant conditions. If the wafer were spinning quickly, rapid vigorous fluid flow would prevent the partially-reduced cobalt ions from accumulating, and nucleation would slow or not occur. On the other hand, once nucleation has occurred, the kinetics of the reducing-agent oxidation and cobalt reduction are quite different. It is believed that cobalt grows on copper in a more rapid, virtually single-step reduction reaction, and fluid convection caused by a high rotational speed enhances mass transfer and deposition rates. Furthermore, the kinetics of the reducing agent (e.g., morpholine borane) are substantially slower on the cobalt surface than on copper. Therefore, during the cobalt growth phase, a different set of chemical (composition and concentration of reducing agents) and physical (e.g., temperature, rotation rate) conditions are desirable.
 Thus, in one aspect of the invention, a method for electroless plating of a cobalt alloy on copper comprises two distinct process phases, nucleation and growth, which are conducted separately under different process conditions, usually with different process chemistry. Typically, a third phase, activation, precedes nucleation.
 The term “activation”, as used broadly herein, means pretreatment to facilitate nucleation. For example, copper typically has a contaminant on it, or a natural oxide, or some chemical that was left there intentionally, such as benzotrizol, which is a copper corrosion inhibitor. A common technique of stripping such materials before nucleation is to expose the surface to a reducing agent, such as dimethylamine borane (DMAB), or to a dilute acid, such as sulfuric acid. Activation includes one or a plurality of operations, depending on the particular pre-conditions and the particular substrate. In a narrow sense, the term “activation” designates a process that makes a surface chemically active for nucleation chemistry. The term “pretreatment” refers generally to processes that prepare a substrate surface for chemical reaction treatment using a thin liquid layer in accordance with the invention. Some pretreatment processes are conducted using thin-liquid-layer techniques in accordance with the invention, while others, for example, spraying, do not use a thin liquid layer. An activation process preceding nucleation is an example of a pretreatment.
 An example of an activation process is removal of a surface contaminant, a natural oxide, or BTA to activate the surface. A typical activation sequence comprises pre-wetting the treatment surface, and then adding a dilute, 1% to 5% sulphuric solution onto the surface. A dilute acid solution reduces the oxide and also tends to corrode it slightly. In addition or alternatively, activation is conducted using a reducing agent, such as DMAB. Other exemplary activating reducing agents are the hypophosphite and glyoxylate ions. Activation is conducted using a thin liquid layer in accordance with one aspect of the invention. Alternatively, activation is conducted using a spray technique or a conventional bath.
 The next phase is nucleation. An exemplary liquid mixture for conducting nucleation of cobalt onto copper comprises: 0.03 mol/l CoCl2.6H2O, 0.06 mol/l citric acid, 0.015 mol/l DMAB. At 60° C. and pH 8.5, the nucleation time is as short as about 2 seconds to 3 seconds. Preferably, nucleation is conducted under quiescent conditions, that is, with minimal or no fluid convection. This is achieved in accordance with the invention by flowing the liquid nucleation mixture into a thin fluid gap to fill the gap, and then decreasing or stopping the flow of solution into the gap.
 An exemplary liquid mixture for conducting the growth phase of electroless plating of cobalt onto copper comprises: 0.03 mol/l CoCl2.6H2O, 0.06 mol/l citric acid, 0.015 mol/l DMAB, 0.03 mol/l ammonium hypophosphite. Addition of hypophosphite ion enhances the subsequent growth rate. Ammonium ion improves the stability of the plating solution, slowing the process and preventing homogeneous reaction. Temperature is typically in a range of from 45° C. to 70° C., preferably about 60° C. The pH is maintained in a range of from 8 to 10, preferably about 9.75.
 The relative amounts of cobalt, boron, and phosphorus in the deposited cobalt alloy is varied by varying the relative concentrations of CoCl2.6H2O, DMAB, and ammonium hypophosphite in the plating-reaction growth-phase solution.
 In one aspect, the process operations of the nucleation and growth phases are conducted separately in accordance with the invention. Preferably, the substantially distinct nucleation and growth phases are conducted in a fluid gap of a microcell device that forms thin liquid layers in accordance with the invention. Nevertheless, it is clear that nucleation-phase processes can be conducted separately from growth-phase processes in accordance with the invention using conventional techniques, such as immersion bath and spraying techniques.
 An advantage of using a microcell or supercell apparatus in accordance with the invention is that the process conditions are closely controlled and dynamically variable. Similarly, the composition of the liquid plating solution is dynamically variable. Another advantage is that many or all of the pretreatment, rinsing, drying and other process operations are conducted in a supercell in accordance with the invention.
 A microcell apparatus and a method using a thin liquid layer in accordance with the invention were utilized to deposit selectively a cobalt-capping layer on top of copper lines, resulting in improved electromigration performance of an integrated circuit. Preparation and processing of the integrated circuit wafers was conducted using four major categories of processes: 1) copper recess and precleaning; 2) cobalt electroless plating; 3) post-cleaning; and 4) rinsing and drying.
 Each wafer was transported to a wafer chuck of a microcell apparatus after CMP copper planarization and barrier (e.g., Ta, TaN) removal, which yielded a surface having a large number of electrically isolated, exposed copper features, such as lines, vias, and pads. The wafer (200 mm diameter) was first wetted with DI water by spraying the water onto the treatment surface while the treating head was in an open, up position away from the surface. Next, the treating head having a substantially flat horizontal treatment surface was lowered to form a thin fluid gap having a width (in the vertical dimension) of about 1.5 mm between the head surface and wafer treatment surface. The wafer was rotated at about 25 rpm, and an etchant was introduced into the thin fluid gap at a rate of about 1 liter per minute until the fluid gap was completely filled (about 10 seconds). Alternatively, the etchant could have been sprayed onto the surface in a continuous fashion at a flow rate of about 1 liter per minute with the wafer rotating at 250 rpm (this alternative uses more chemical). We have found that it was desirable to recess the exposed copper structures on each of the wafers approximately 100 Å to 300 Å below the plane of the dielectric to improve the overall performance of the resulting circuit structures. It is believed that by performing this controlled recess: 1) damaged surface copper left over from the CMP process was removed, thereby improving the activity of the copper for nucleation of cobalt onto the surface; and 2) by plating into a recessed structure, lateral growth of the capping layer was prevented (by the confinement of the trench wall), thereby reducing or eliminating the encroachment of adjacent conductive lines. Generally, the copper was recessed with an ambient-temperature reactive etching mixture of hydrogen peroxide (from 1% to 5%) and glycine (from 0.2% to 2%). The pH of the solution was changed to a slightly alkaline pH (about 8 to 10) by the addition of an alkaline agent, such as tetramethylammonium hydroxide (TMAH), for optimum rate control and reproducibility. This mixture resulted in a conformal recessing process (i.e., relatively flat etch profile) and a slow, controllable etching rate (50 Å/min to 300 Å/min), independent of the size of the feature. Target amount of metal recess was in a range of about 100 Å to 300 Å. After the desired amount of time (etched thickness), the etchant was removed from the fluid gap area by flushing the inlet line and fluid gap with DI water.
 Generally, after these etching and rinsing operations, the wafer was megasonically cleaned with a dilute solution containing a copper complexing agent (e.g., citric acid, EDTA). A slow rotation rate (10 rpm to 25 rpm) combined with about 125 Watts of megasonic energy at about 0.85 MHz was found to be suitable. This cleaning helped to remove spurious particles and extract copper (and other) metal ions that might have been ion-exchanged with the surface dielectric layers during the formation of metal ions in the etching step. After this clean, the wafer was again rinsed and was ready for selective electroless cobalt plating.
 The selective electroless cobalt plating is described here with reference to FIG. 14. About 3 liters of cobalt plating solution stored in a heated tank 924 was continuously circulated and filtered 942 at 1 liter per minute. Generally, the plating solution contained a mixture of 30 g/L COCl2*6H2O, 54 g/L NH4Cl, 57 g/l citric acid monohydrate, and about 625 g/L of 25%/w TMAH. The solution was heated to 72° C., and the solution pH was 9.75 at that temperature. An in-line flow meter 944 measured flow rate upstream from a three-way valve 946 that was used either to direct plating solution back (recirculation stream 948) to-heated storage tank 924, or to direct the plating solution (undiverted stream 950) to microcell reactor head 922. To feed the plating solution to the microcell's thin fluid gap 934, three-way valve 946 was opened and fluid passed into line 950 at 1 liter/minute past a tee 952 used for other purposes (e.g., later flushing of chemicals). A tee 954 served as an injection point for the introduction of reducing agent from a second liquid source 926. The reducing agent solution (80 g/L DMAB, 15 g/L ammonium hypophosphite, and 25 ml/L 25%/w TMAH) was introduced into line 958 at a rate of 85 ml/min at the same time as the reactant from the heated storage tank. Downstream of tee 954, inline mixer 960 provided mixing of the two liquid solutions to form a reaction mixture in inlet line 962.
 To insure complete gap filling, the wafer was rotated at 75 rpm while the liquid reaction mixture was injected for about 7 seconds (200 mm wafer, 1.5 mm gap). Treating head 922 (containing imbedded flow lines and fittings to incoming fluid lines) was made of titanium, had a circular diameter of approximately 200 mm, and had a vertical thickness of approximately 1 inch. This head was attached to a thermal mass and temperature control block (made of highly conductive aluminum) using thermal contact paste. The block was about 65 mm thick and 220 mm in diameter. An electric heating coil was attached to the top of this heater block. A thermocouple temperature probe was embedded in titanium treating head 922 and was used with a feedback controller to maintain the head temperature at 70° C. The treating head comprised a showerhead-type outlet for flowing liquid into the thin fluid gap. The outlet comprised about 50 outlet holes arranged in a 2-inch diameter circle, each of the holes having a diameter of about 0.8 mm. The lower head surface of titanium treating head 922 adjacent to fluid gap 934 was coated with a PTFE film (spray coated, about 0.1 mm thick).
 After the liquid reaction mixture filled the fluid gap without any entrapped bubbles, the liquid flow into the fluid gap and the rotation of the wafer were completely stopped (i.e., no flow, 0 rpm) for a period of 20 seconds. During this time, the solution was quiescent, a desirable condition for nucleation during initiation of the plating process at the copper surface. After this nucleation period, the wafer was generally rotated at a very slow rate (2 rpm to 10 rpm) until completion of the plating growth phase (to the desired thickness). This slow rotation rate was found to improve azmuthal uniformity of the plating process. The film growth rate during the growth phase was approximately 250 Å/min (actual value depended strongly on exact pH and temperature).
 At the end of plating, DI water was injected into tee 952. The preliminary rinsing of the wafer and head was accomplished by rotating the wafer at 150 rpm with a DI flow rate of 1 liter/m for about one minute.
 The post-plating cleaning processes involved the injection of a treatment chemistry suitable for removing spurious cobalt particles from the surface and metal ions from the dielectric (ion exchange removal). The solution combined a buffered pH and strong complexing agents with a weak etching character. A suitable solution for cobalt-cap-layer treatment was a 10-1 dilution of the proprietary cleaning solution ESC-784 (ESC Incorporated, Bethlehem, Pa.). This solution was combined with a simultaneous megasonic treatment (125 W, 0.85 MHz, 12 RPM, 1 liter/min flow, 1 minute treatment), and was followed by a DI rinse of the liquid tubing lines, treating head, fluid gap, and wafer (150 rpm, 1 l/min). The rinse water was ejected from the gap by introducing pressurized nitrogen (flow rate 700 cc/min STP) into the gap and increasing the rotation rate to about 300 rpm. With the wafer still rotating, the head was lifted away from the wafer and DI rinse water was sprayed onto the wafer's surfaces. The treating head was then tilted to allow excess water to drain from it, and the treating head was allowed to dry.
 TEM imaging of the wafers showed that the film thickness of the cobalt film on top of the copper was generally about 300 Å, and was confined within the trenches in the dielectric material of the wafer. Cobalt films of 100 Å thickness were also deposited on copper. The cobalt films were conformal, covering the copper lines. SEM was conducted of a copper damascence comb test, patterned and capped with cobalt. Results of SEM showed that the underlying copper lines were completely and selectively coated with cobalt, without spurious material created between the lines.
 Various performance data show good electronic properties on integrated circuit wafers fabricated in accordance with the invention. A line leakage comparison for conductive copper lines capped with cobalt showed no significant difference in the performance of the recessed and capped features compared to the controls. In contrast, non-recessed features showed considerably higher leakage currents, presumably because of the average closer distance between the edge of the cobalt-capped areas and the edge of the lines (encroachment). FIG. 15 shows the results of electromigration (EM) tests comparing EM lifetime of Co-capped Cu with the EM lifetime in baseline wafers having Cu lines without Co-capping. Control baseline wafers did not have a Co-capping layer, but were otherwise processed in the same manner as the cobalt-processed wafers. The average EM lifetime of control wafers is about 50 hours. In contrast, wafers plated electrolessly with Co in accordance with the invention showed significant improvement of EM lifetime (at least 10× increase). In the case of wafers having Co-capping of 100 Å thickness, there was no failure during 500 hours of the EM test time. There was one failure out of 16 samples that had 300 Å thick Co-capping. These results show that electroless capping with cobalt in accordance with the invention is very effective in improving EM lifetime of narrow Cu-interconnects.
 The particular systems, designs, methods and compositions described herein are intended to illustrate the functionality and versatility of the invention, but should not be construed to be limited to those particular embodiments. Systems and methods in accordance with the invention are useful in a wide variety of circumstances and applications to conduct a liquid chemical reaction treatment and other liquid-phase treatments performed on an integrated circuit substrate. It is evident that those skilled in the art may now make numerous uses and modifications of the specific embodiments described, without departing from the inventive concepts. It is also evident that the steps recited may, in some instances, be performed in a different order; or equivalent structures and processes may be substituted for the structures and processes described. Since certain changes may be made in the above systems and methods without departing from the scope of the invention, it is intended that all subject matter contained in the above description or shown in the accompanying drawings be interpreted as illustrative and not in a limiting sense. Consequently, the invention is to be construed as embracing each and every novel feature and novel combination of features present in or inherently possessed by the systems, methods and compositions described in the claims below and by their equivalents.