DE102016206769B3 - Method for structure-dependent filling of depressions - Google Patents

Abstract

The present invention relates to a method for the structure-dependent filling of depressions (1, 2, 3) in a dielectric substrate, in which at least two depressions of different widths or different diameters (1, 2, 3) are introduced into the substrate in one step. Thereafter, a first filling layer (5) of a first material is applied to the first layer (4), walls of the recesses (1, 2, 3) and webs, wherein a thickness (d film) of the applied first filling layer (5) at least half a width corresponds to that of the recesses (1) having the smallest width or the smallest diameter (ITrench1), so that this recess (1) is completely filled by the first filling layer (5). Subsequently, a second filling layer (6) made of a second material is applied, so that all recesses (1, 2, 3) are filled and a substrate surface (7) is covered with the second material. Finally, in step v), the first material and the second material are removed from the substrate surface (7) by chemical-mechanical planarization until the substrate surface (7) has been exposed.

Description

The present invention relates to a method for structure-dependent filling of depressions in a dielectric substrate.

In the manufacture of integrated circuits (IC) substrates must be structured and provided with tracks. A common process for creating printed circuit traces and vertical interconnects, also referred to as vertical vias or "vias" (vertical interconnect access), is the so-called dual damascene method. However, this method is increasingly technological limits set and the ever-increasing demands for ultra-thin and at the same time compliant, closed barrier, adhesion and germ layers can be met only with great effort. With the dual damascene process, therefore, a realization of different conductivities within a management level is only possible to a limited extent.

With current structure sizes up to 28 nm, a structure-selective adaptation of a material or material used to provide different conductivities and an improvement in the electromigration resistance in a management level is not yet necessary. However, this requirement is developing at ever smaller technology nodes to z. B. to realize different conductivities.

• a) Unterschiedliche Leitfähigkeiten können durch das Layout künstlich erzeugt werden. Hohe Leitfähigkeiten können über entsprechend breite Leiterbahnen realisiert werden. Dies führt aber zu einem Platzproblem bei kleinen Technologieknoten.
• b) Je nach Layout kann eine Leiterbahn künstlich verlängert werden. Dies führt zu einem Platzproblem bei kleinen Technologieknoten.
• c) Bei kleinen Strukturen können andere Materialien bzw. Werkstoffe mit anderen Eigenschaften genutzt werden. Dies hat aber den Nachteil, dass die gleichen Eigenschaften nicht optimal für große Strukturbreiten sind.

In a conventional process flow, ie in the dual damascene process, there are the following possible solutions, in which compromises have to be made:

• a) Different conductivities can be artificially generated by the layout. High conductivities can be realized via correspondingly wide strip conductors. However, this leads to a space problem with small technology nodes.
• b) Depending on the layout, a trace can be artificially extended. This leads to a space problem with small technology nodes.
• c) For small structures, other materials or materials with different properties can be used. However, this has the disadvantage that the same properties are not optimal for large feature sizes.

For example, the document discloses US 2015/0311113 A1 a training of different conductivities by different trench depths. Again, this is very difficult to realize experimentally, since a block-wise processing at different structure sizes is very complex, since at least doubling the number of required process steps compared to conventional methods.

From the publication US 2007/0144628 A1 For example, a method of forming a plurality of electrically conductive structures on a substrate is known. In this case, recesses in the substrate are filled by electrically conductive coatings. Also the publication US 2006/0091551 A1 shows a similar method for forming electrically conductive connections on a substrate.

The present invention is therefore based on the object to propose a method for the structure-dependent filling of wells, which avoids the disadvantages mentioned, with which so conductor tracks of different conductivities in a substrate can be produced with as few process steps.

This object is achieved by a method according to claim 1. Advantageous embodiments and further developments are described in the dependent claims.

In a method for the structure-dependent filling of depressions in a dielectric substrate, at least two depressions of different widths or different diameters are introduced into the substrate in one step. In the subsequent step, a first filling layer of a first material is applied to floors, walls of the recesses and webs (which are located, for example, between the recesses). A thickness of this first filling layer corresponds to at least one half width of that of the depressions, which has the smallest width or the smallest diameter of the introduced depressions, so that this depression is completely filled by the first filling layer. Subsequently, in a further step, a second filling layer of a second material is applied, so that all depressions are filled and a substrate surface is covered with the second material. Finally, by chemical mechanical planarization (CMP), the first material and the second material are removed from the substrate surface until the substrate surface has been exposed.

Due to a decoupling of a metallization of small and larger structures, the described method allows the use of different materials or material mixtures for different conductor tracks. A comparatively small number of steps makes it possible to simplify the entire process more easily. In addition, a number of internal interfaces are reduced. This also allows a more targeted adjustment of electromigration and an increase in reliability. The depressions are usually used as a template for printed conductors of a conductor track plane, in particular vertical vias should be produced in the wells.

The first material and the second material are typically metals. As a first material or as a second material, titanium (Ti), cobalt (Co), ruthenium (Ru), tungsten (W), aluminum (Al), nickel (Ni), Mn (manganese), tantalum (Ta), chromium ( Cr), copper (Cu) or silver (Ag). Copper may be preferred for so-called "power lines", while cobalt can be used for dimensionally smaller "clock lines". Alternatively or additionally, for the first material or for the second material, however, it is also possible to use compounds of the abovementioned materials, in particular silicides, germanides, preferably CoGe ₂ , Cu ₃ Ge, NiGe, nitrides, preferably CoN, TaN, TiN, aluminides, preferably CuAl or Co Al, or manganides. Of course, it is also possible to use an alloy of the abovementioned materials, preferably a Co-W compound, or the abovementioned materials, alloys and compounds can be used in any combination as the first material and as the second material, provided that the first material of the second Material is different.

It can be provided that the first filling layer and / or the second filling layer is or are formed as a homogeneous layer or as homogeneous layers. A homogeneous layer is to be understood as meaning, in particular, a layer which completely covers the surface on which this layer is deposited, that is to say a closed layer which typically has an always constant thickness. By forming a homogeneous layer, unwanted plated-through holes, for example through holes in the layer, are avoided.

The depressions should each be formed with the same or identical depth. By the introduction into the substrate takes place with the same depth, the conductivity of the conductor tracks produced depends only on the diameter at circular recesses or the width and the material used to fill the recess or the material mixture used for filling. Thus, the desired conductivity is easier to set.

Before applying the first filling layer to the bottoms and walls of the depressions, a first intermediate layer can be applied to the bottoms and the wall on which the first filling layer is deposited. This first intermediate layer preferably serves as a barrier layer and / or as a seed layer. Typically, the first intermediate layer by atomic layer deposition (ALD), chemical vapor deposition (CVD) and / or physical vapor deposition (PVD), so-called "electroless plating", in which without the use of an external a metal deposition takes place electric current, and / or applied an electrochemical deposition process.

The first filling layer and / or the second filling layer can be applied by immersion in a copper salt bath, preferably with a copper electrolyte, atomic layer deposition, chemical vapor deposition and / or physical vapor deposition and / or an electrochemical deposition process.

It is also provided that the first filler layer is aftertreated after application by nitrogen. The second filling layer can also be post-treated after application by nitrogen. The first filling layer and / or the second filling layer may also alternatively or additionally be obtained by a heat treatment such as annealing and / or a plasma method, i. H. a plasma treatment, aftertreated. By the heat treatment, a surface roughness of the first filler layer can be reduced. The nitrogen treatment provides better corrosion resistance.

On the first filling layer and / or on the second filling layer, a second intermediate layer for a simplified application of the second filling layer can be applied. This intermediate layer is preferably applied by means of atomic layer deposition, chemical vapor deposition and / or physical vapor deposition, electroless plating and / or an electrochemical deposition process.

The first intermediate layer and the second intermediate layer (if the second intermediate layer is provided in the layer system to be produced) may comprise a barrier layer, an adhesion promoter layer and a seed layer. The barrier layer is preferably made of tantalum nitride, the adhesion promoter layer is preferably formed of tantalum. Copper is preferably used for the seed layer.

The dielectric substrate should be formed with at least two inter-level dielectric (ILD) dielectric layers. Preferably, a diffusion barrier layer is provided between the two dielectric layers.

The diffusion barrier layer is hereby penetrated by the depressions, which takes place when the depressions are introduced into the substrate. The dielectric layers may have a same or identical thickness.

The introduction of the depressions can take place with one of the methods customary in semiconductor processing, that is to say, for example, application of a mask and a subsequent etching step. Even when applying the first intermediate layer, the substrate should be covered with a mask except for the area to be coated.

Embodiments of the invention are illustrated in the drawings and are described below with reference to the 1 and 2 explained.

Show it:

1 a sectional view of a dielectric substrate having a plurality of vertical vias and

2 a 1 corresponding view of the substrate with planarized surface.

1 shows a sectional view of a dielectric substrate ILD. In further exemplary embodiments, it is also possible to use a multilayer substrate, wherein a diffusion barrier layer of silicon oxide can also be arranged between the layers. On an underside of the dielectric substrate ILD is an electrical contact area 10 arranged. The dielectric substrate ILD has already been made with pits 1 . 2 and 3 Mistake. This was done by applying a mask and etching until reaching the electrical contact area 10 , The dielectric substrate ILD is made of silicon dioxide in the illustrated embodiment.

Each of the wells 1 . 2 and 3 has a different width, but all are equally deep, ie they extend to the respective contact area 10 , The different width of the wells 1 . 2 and 3 Expresses itself also in the fact that the widths at the contacting 10 I _Via1 and I _Via2 the wells 1 and 2 as in 1 are equal in size but smaller than the corresponding width I _{Via3 of} the recess 3 , When deepening 1 corresponds to the upper width of _Trench1 just the width at the contact I _Via1 , while deepening 2 the upper width I _{Trench2 is} just twice as large as the width at the contact I _Via2 . Also at the recess 3 the upper width I _{Trench3 is} significantly larger than the width I _Via3 .

While the in 1 left-hand recess 1 has the smallest width and runs as a constant diameter passage through the substrate ILD, the in 1 right-hand recess 3 the largest width which is constant in an upper part of the substrate ILD and continuously decreases in a lower portion of the dielectric substrate ILD to the contact area 10 runs. The middle depression 2 Finally, it has a greater width in the upper area of the dielectric substrate ILD than in the lower area of the substrate ILD.

A width of the contact areas 10 the wells 1 . 2 . 3 is different in size in the illustrated embodiment, but in other embodiments, the width can also be the same size. On a floor, walls and ridges of each of the pits 1 . 2 . 3 is a germ layer 4 deposited by means of atomic layer deposition on which a coating to be applied adheres and with the adaptation of the lattice constants of the materials to be joined and thus improved adhesion can be achieved.

On this germ layer 4 was in the illustrated embodiment, a first filling layer 5 of cobalt deposited by chemical vapor deposition. This can be done in the usual way by applying a mask, coating and removing the mask. The first filling layer 5 was deposited in a thickness d _film , which is at least half the width I _{Trench1 of} the recess 1 corresponds to the smallest width of all wells 1 . 2 . 3 having. Thus, the depression became 1 completely through the first filling layer 5 refilled.

The first filling layer 5 was then subjected to a heat treatment to reduce a surface roughness and treated with nitrogen to increase the corrosion resistance. Finally, a second layer 8th by means of physical vapor deposition on the surface of the first filling layer 5 deposited as a layer system is formed of a barrier layer of tantalum nitride, a primer layer of tantalum and a seed layer of copper. On this second layer 8th became a second filling layer 6 of copper by immersion in a copper electrolyte salt bath and electrochemical deposition applied.

A substrate surface 7 is thus complete with the first filling layer 5 and the second filling layer 6 covered. Likewise, all pits are 1 . 2 . 3 either through the first filling layer 5 (Deepening 1 ) or a combination of first filling layer 5 and second filling layer 6 (wells 2 and 3 ) filled. Both are the first filling layer 5 as well as the second filling layer 6 formed as a homogeneous, ie closed and hole-free layers, so that the only electrical connection between the top and bottom over the in the wells 1 . 2 . 3 introduced metals of the two filling layers 5 and 6 he follows.

By forming and providing structure-size-dependent properties required within a cable path level, a simple structuring of the substrate is achieved, which would only be achievable with considerably more effort using conventional methods.

In 2 is in one 1 corresponding view the in 1 already shown substrate shown. Recurring features in this figure are given the same reference numerals as in FIG 1 Mistake. In contrast to the processing state in 1 However, a chemical-mechanical planarization was carried out in which the first material of the first filling layer 5 and the second material of the second filling layer 6 from the substrate surface 7 were removed, leaving this substrate surface 7 now exposed. On the substrate surface 7 Now further layers can be arranged, for example, further interconnects or an encapsulation.

Claims (10)

Method for structure-dependent filling of recesses ( 1 . 2 . 3 ) in a dielectric substrate, in which, in a step i), at least two recesses of different widths or different diameters ( 1 . 2 . 3 ) are introduced into the substrate and ii) a first filling layer ( 5 ) of a first material on floors, walls of the recesses ( 1 . 2 . 3 ) and webs is applied, wherein a thickness (d _film ) of the applied first filler layer ( 5 ) at least half a width of those of the depressions ( 1 ), which has the smallest width or the smallest diameter (I _Trench1 ), so that this depression ( 1 ) through the first filling layer ( 5 ) is completely filled, wherein the first filling layer ( 5 ) is aftertreated by nitrogen application after application, then in step iii) a second filler layer ( 6 ) is applied from a second material, so that all depressions ( 1 . 2 . 3 ) are filled and a substrate surface ( 7 ) is covered with the second material, and finally in step iv) by chemical-mechanical planarization of the first material and the second material of the substrate surface ( 7 ) are removed until the substrate surface ( 7 ) has been exposed. A method according to claim 1, characterized in that as a first material or as a second material titanium, cobalt, ruthenium, tungsten, aluminum, nickel, manganese, chromium, copper or silver, a compound of said elements, preferably a silicide, a germanide, especially preferably CoGe ₂ , Cu ₃ Ge, NiGe, a nitride, particularly preferably CoN, TaN, TiN, an aluminide, particularly preferably CuAl, CoAl, or a manganide, or an alloy of said materials, particularly preferably a cobalt-tungsten alloy, is used. Method according to claim 1 or claim 2, characterized in that the first filling layer ( 5 ) and / or the second filling layer ( 6 ) is formed as a homogeneous layer (s). Method according to one of the preceding claims, characterized in that the depressions ( 1 . 2 . 3 ) are each formed with a same depth. Method according to one of the preceding claims, characterized in that before the application of the first filling layer ( 5 ) on the bottoms and walls of the depressions ( 1 . 2 . 3 ) at least a first intermediate layer ( 4 ) is applied, which preferably serves as a barrier layer and / or seed layer, wherein the first intermediate layer ( 4 ) by atomic layer deposition, chemical and / or physical vapor deposition, electroless plating or an electrochemical deposition process. Method according to one of the preceding claims, characterized in that the first filling layer ( 5 ) and / or the second filling layer ( 6 ) by immersion in a metal salt bath, atomic layer deposition, chemical and / or physical vapor deposition, electroless plating or an electrochemical deposition process. Method according to one of the preceding claims, characterized in that the first filling layer ( 5 ) and / or the second filling layer ( 6 ) is aftertreated by a heat treatment and / or a plasma process. Method according to one of the preceding claims, characterized in that on the first filling layer ( 5 ) a second intermediate layer ( 8th ) for easier application of the second filler layer ( 6 ), wherein the second intermediate layer is preferably applied by means of atomic layer deposition, chemical and / or physical vapor deposition, electroless plating or an electrochemical deposition process. Method according to the preceding claim, characterized in that the first intermediate layer ( 4 ) and / or the second intermediate layer ( 8th ) is formed by a barrier layer, which is preferably formed from tantalum nitride, an adhesion promoter layer, which is preferably formed from tantalum, and a seed layer, which is preferably formed from copper. Method according to one of the preceding claims, characterized in that a dielectric substrate is used, which with at least two dielectric layers (ILD) is formed, wherein preferably a diffusion barrier layer ( 9 ) is arranged between the at least two dielectric layers (ILD) and wherein the diffusion barrier layer ( 9 ) from the depressions ( 1 . 2 . 3 ) is broken in step i).

Images