Sidewall coverage for copper damascene filling

[0002] Continuation in part of application Ser. No. 09/358,983 which was filed Jul. 22, 1999.

[0003] The invention relates to the general field of integrated circuit manufacture with particular reference to techniques for filling small opening with metal, for example copper in a damascene structure.

[0004] During the course of manufacturing an integrated circuit, the need frequently arises to fill an opening in the form of a hole or trench (generally in the surface of a dielectric layer) with material such as tungsten or copper. A particular example of this is the well-known damascene process in which conducting lines are formed that are flush with the surface rather than lying on top of it.

[0005] We illustrate, in FIG. 1, a particular problem which arises when a hole or trench needs to be filled with copper. This problem becomes increasingly more acute as the diameter or width of the opening gets to be less than about 0.3 microns. Before the opening can be filled using a high-volume technique such as electroplating, it is necessary to lay down a seed layer of copper. In FIG. 1 we show a cross-section of a portion of dielectric (or some other material) layer 11 in which opening 15 has been formed. Layer 12 is a layer of field oxide that is normally present although not directly relevant to the practice on the present invention.

[0006] The problem mentioned above arises with the deposition of seed layer 13 (usually, but not necessarily) of copper. The most widely used processes for depositing the seed layer are vacuum evaporation and sputtering (known collectively as PVD or physical vapor deposition). Because of shadowing effects, there is a tendency for more material to build up near the mouth of the opening than lower down, giving the deposited seed layer the profile shown in FIG. 1. In particular, there can be substantial overhang of the seed layer, as pointed to by arrows 14, near the mouth of the opening.

[0007] When layer 13 is later built up, typically by electroplating, so as to fully fill opening 15, material at the edges of the overhang come together before the hole can be fully filled, resulting in the trapping of a void within the copper plug (sometimes called the key hole effect).

[0008] An obvious approach to dealing with this problem is to reduce the amount of overhang to the point that void trapping does not occur. In the prior art this is done either by limiting the amount of copper deposited in the first place or by etching the seed layer back using conventional chemical means, such as wet or dry etching.

[0009] Simply reducing the thickness of the seed layer, while reducing the possibility of void formation, introduces a new problem which is illustrated in FIG. 2. Shown there is seed layer 23, having reduced thickness relative to seed layer 13 of FIG. 1. However, if the overhang 24 is sufficiently reduced, bare spots such as 25 begin to appear on the side walls of the opening. The presence of such bare spots then has disastrous consequences for the subsequent hole filling procedure since multiple voids and poorly adherent areas get formed. The present invention shows how a seed layer having little or no overhang, while at the same time having sufficient thickness to fully cover the side walls, may be formed.

[0010] Effect on contact resistance between wiring layers

[0011] We note here that, because of the problems outlined above, it was necessary to limit trench depths so as to keep their aspect ratio (as seen in cross-section) to less than about 6:1, otherwise there was a danger of voids forming when they were filled with metal. This in turn meant that via holes extending downwards from the trench bottom to the next wiring level (i.e. dual damascene structures) had to be correspondingly deeper (i.e. their aspect ratio would typically be at least 6:1). As a result, the series resistance of the via (i.e. wiring level to wiring level contact resistance), when filled with metal, would be larger than desired. Typically, a contact resistance less than 1 ohm for a via size of 0.2 microns could occur. In FIG. 6 we show a typical dual damascene structure of the prior art in which the depth of trench 61 is T1 and the depth of via 62 is V1. It follows that any increase in the ratio T1/V1 will reduce the wire-to-wire contact resistance.

[0012] A routine search of the prior art was conducted but no references were found that teach the solution described by the present invention. Several of these references were, however, of interest. For sample, Crank (U.S. Pat. No. 5,316,974) limits the seed layer to the bottom of the trench so that the filler plug grows (by electroplating) from the bottom up and not from the vertical sides, thereby avoiding void formation.

[0013] Zhao et al. (U.S. Pat. No. 5,674,787) form the seed layer by exposing the conductor at the bottom of the hole and then dipping it in a solution that deposits a thin copper layer by displacement. A plug is then grown on this seed layer using an electroless process.

[0014] Venkatraman (U.S. Pat. No. 5,677,244) dopes aluminum with copper by first laying down an agglomerated copper film, then depositing the aluminum copper then heating so as to diffuse the copper islands throughout the aluminum.

[0015] Venkatraman et al. (U.S. Pat. No. 5,814,557) describe a process for filling the trench/hole of a damascene structure by depositing two different conductors one after the other.

[0016] It has been an object of the present invention to provide a process for filling a trench or hole at the surface of an integrated circuit without trapping voids or leaving bare spots on the side walls.

[0017] Another object of the invention has been to provide an apparatus in which to implement the process of the present invention.

[0018] A further object of the invention has been to provide a process for the formation of damascene wiring that is free of trapped voids (key hole effect).

[0019] A still further object of the invention has been that said process and apparatus be fully compatible with current techniques in use for the manufacture of semiconductor integrated circuits.

[0020] These objects have been achieved by first depositing a seed layer in the hole or trench by means of PVD, in the usual manner. This is then followed by a sputter etching step which removes any overhang of this seed layer at the mouth of the trench or hole. It is in general preferable to perform the deposition and etching of the seed layer in the same apparatus without breaking vacuum between operations. When the sputter etch conditions specified by the invention are followed all overhang by the seed layer at the mouth of the trench or hole is essentially eliminated without the introduction of any poorly adhering or bare spots on the side walls.

[0021] This technique can be further improved by several refinements including multiple, sputter etching and seed layer deposition steps, increasing the T/V ratio discussed above, and using sputtering under varying conditions of pressure and voltage both to form the seed layer as well as to flatten it.

[0022] FIG. 1 illustrates how a seed layer, deposited for the purpose of lining a trench or hole, can exhibit considerable overhang near the mouth of the opening.

[0023] FIG. 2 illustrates how attempts of the prior art to solve this problem have led to new problems in the form of bare spots on the side walls.

[0024] FIGS. 3aand 3bshow two embodiments of the apparatus used for implementing the present invention.

[0025] FIG. 4 shows how the profile of a seed layer can exhibit little or no overhang after processing according to the present invention

[0026] FIG. 5 illustrates how a hole or trench may be over filled without trapping of any voids.

[0027] FIG. 6 shows a dual damascene structure in which limitations of the prior art cause a higher than desired wire-to-wire contact resistance.

[0028] FIG. 7 is the structure of FIG. 6 with said contact resistance has been reduced.

[0029] FIG. 8 illustrates an apparatus in which sputtering may be used to both deposit and etch a seed layer.

[0030] The process of the present invention begins, as in the prior art, with the formation of an opening such as a hole or trench (such as 15 in FIG. 1) in the surface of an integrated circuit. Most commonly, said surface will be that of a dielectric layer, but that is not a requirement of the invention. Typically the trench width used will be between about 0.1 and 15 microns, the hole diameter between about 0.1 and 0.5 microns, and the depth of the opening will be between about 0.4 and 1 microns.

[0031] Next, if a copper damascene process is being used, a barrier layer of a material such as tantalum, tantalum nitride, titanium nitride, or tungsten nitride is laid down to a thickness between about 100 and 500 Angstroms. Whether or not a barrier layer is used, a seed layer of metal (specifically copper if a damascene process is being used) is next laid down to cover the inside surfaces of the trench or hole. The thickness of this seed layer is between about 800 and 2,500 Angstroms and it is deposited by means of PVD (sputtering or vacuum evaporation).

[0032] Once the seed layer is in place, a process step not currently practiced in the prior art is introduced. This is the removal of a certain amount of the seed layer (typically between about 100 and 500 Angstroms) by means of sputter etching. This latter step is performed using argon at a pressure between about 0.1 and 2 mtorr, at a power level between about 150 and 450 watts. Sputter etching may be effected by using either DC or RF.

[0033] Because of shadowing effects, material from the seed layer is preferentially removed from the overhanging portion at the mouth of the trench or hole. For best results, it is important that the sputtering conditions specified above are used. Selective removal of the overhang will not be achieved if, for example, the pressure used for sputter etching is too high.

[0034] While the invention will still be effective if the PVD and sputter etching steps are performed in separate chambers we have preferred to use a single chamber for both of these processes because of the resulting improved throughput. An example of the apparatus in which this in-situ deposition and etching are performed is shown in FIG. 3a.Evacuable chamber 31 is provided with a pumping port 32 as well as adjustable inlet 36 for the controlled admission of a sputtering gas (generally, but not necessarily, argon). In the example shown in FIG. 3a,the filler material (copper if a damascene process is being used) is deposited in a downwards direction from a suitable source such as tungsten filament 35. To ensure good coverage by the deposited metal, a plurality of such sources, typically arranged in a ring, are often used. An electrode 33, having a flat horizontal surface and located near the bottom of the chamber, is used to support the integrated circuit wafer(s) during metal deposition and also serves as the electrode to which power for the sputter etching (symbolized as 34 in the figure) is delivered. Since the wafers are supported from below, it is not required that they be clamped to the electrode.

[0035] In an alternative embodiment of the apparatus, the arrangement shown in FIG. 3bwas used. Evacuable chamber 31 is provided with a pumping port 32 as well as adjustable inlet 36 for the controlled admission of a sputtering gas (generally, but not necessarily, argon). In the example shown in FIG. 3b,the filler material (copper if a damascene process is being used) is deposited in an upwards direction from a suitable source such as crucible 65. To ensure good coverage by the deposited metal, a plurality of such sources, typically arranged in a ring, are often used. An electrode 33, having a flat horizontal surface and located near the top of the chamber, is used to support the integrated circuit wafer(s) during metal deposition and also serves as the electrode to which power for the sputter etching (symbolized as 34 in the figure) is delivered. In this embodiment, it is necessary to clamp the wafers to the electrode.

[0036] Regardless of the exact manner in which layer 13 (FIG. 1) was deposited, the introduction of the extra sputter etching step (assuming the conditions that we have specified are followed) has the effect of preferentially removing metal from the overhang, resulting in the seed layer profile shown in FIG. 4. As can be seen, seed layer 43 now has an outer surface that closely parallels the contours of the original trench or hole. When this is the case, the opening may be filled, using a deposition technique such as electroplating, and then over-filled, giving it the profile shown in FIG. 5, where layer 53 represents the filler layer. Note that no voids have been trapped within 53 and that no bare spots are present at the trench walls.

[0037] Once the structure shown in FIG. 5 has been obtained, formation of the damascene structure may be completed by planarizing the surface of 53. Of several available techniques CMP (chemical mechanical polishing) is the most widely used for this purpose.

[0038] The basic process described above may be further improved by the addition of several refinements:

[0039] (1) After most of the overhang by the seed layer at the mouth of a cavity has been removed, as described above, a second seed layer deposition is undertaken followed by a second sputter etching step. These additional steps further improve the profile of the seed layer at the mouth of the cavity. Typically, the initial (first) thickness of the seed layer would be between about 800 and 2,500 Angstroms. The amount of seed layer removed during the first sputter etching step would be between about 400 and 1,000 Angstroms while the amount of seed layer removed during the second sputter etching step would be between about 400 and 1,000 Angstroms.

[0040] (2) By taking advantage of the fact that deeper cavities can now be properly filled, it becomes possible to increase the depth of the trench portion of a dual damascene structure which allows for a reduction in the depth of the via portion. This, in turn, reduces wire-to-wire contact resistance to values as low as about 0.5 ohms for an 0.2 micron via, as illustrated in FIG. 7 where the ratio T2/V2 is seen to have been increased relative to T1/V1 in FIG. 6.

[0041] (3) By arranging to use a single chamber, as illustrated in FIG. 8, for both the deposition and the sputter etching of the seed layer, a cleaner process can be achieved since there is need for only a single pump down.

[0042] (4) As a further refinement of process (3), the pressure of the sputtering gas (usually argon) is arranged to be relatively high (between about 0.01 and 100 mtorr) during the deposition step and then, at the same time that the voltage application point and magnitude (and hence the power) are changed, the pressure is arranged to be relatively low.

[0043] (5) In what is partly a reversal of (4) above, a sufficient amount of the seed layer is removed to ensure that no overhang remains, following which the amount removed (typically between about 400 and 1,000 Angstroms) is replaced through sputter deposition. By using a relatively high pressure for the latter step (between about 10 and 90 mtorr), the non re-appearance of the overhang is ensured.

[0044] While the invention has been particularly shown and described with reference to the preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made without departing from the spirit and scope of the invention.