Semiconductor device manufacturing: process – Coating with electrically or thermally conductive material – To form ohmic contact to semiconductive material
Reexamination Certificate
1999-06-09
2002-06-18
Nelms, David (Department: 2818)
Semiconductor device manufacturing: process
Coating with electrically or thermally conductive material
To form ohmic contact to semiconductive material
C438S582000
Reexamination Certificate
active
06406997
ABSTRACT:
FIELD OF THE INVENTION
This invention relates generally to chromium films, and more specifically to the use of chromium films to reduce electromigration in semiconductor and related devices.
BACKGROUND OF THE INVENTION
Integrated circuits get smaller and smaller. This makes it possible to place more and more transistors on a given size of an integrated circuit. This requires that connecting lines between transistor and other devices on the silicon chip get smaller and smaller. High current densities are used in such conductors.
Currents in metals are usually carried by electrons. However, if current densities are in the 10
6
A/cm
2
range, and the temperature is high, metal ions are also moving. This phenomena is called electromigration and changes the geometry of the conductors. Hillocs and cracks form and lead to the degradation and failure of electrical conductors. Ions move usually in the same direction as electrons, because the momentum transfer from electrons to ions is responsible for the movement of the ions.
Surface diffusion and surface coatings influence electromigration. Surface coating as protection against failure is a subject of controversy. Opinions range from “no-effect” to “cure-all” according to a review by F. M. D'Heurly and R. Rosenberg (1). These authors suggest that the surface must be considered mainly for the effect of vacancy movement on the surface. Clearly, overlayers produce specific vacancy concentrations at the interface.
Schwarz (2) in his review of electromigration in interconnects and contacts gives a summary of results of such studies. Aluminum strips covered with silicon dioxide showed an increase by a factor of 3 in the Mean Time to Failure (MTF) (3), or by a factor of 23 by first anodizing films and then adding a porous layer (4). Sputter conditions of the silicon dioxide influenced the MTF (5). Al-Cu conductors (6) with 1.5 &mgr;m of glass had a longer lifetime than those without glass. The glass overlayer produced changes in the failure mechanism.
A brief review included in a recent report on electron microscopy studies of electromigration defects (7) shows that there is now an increasing interest in electromigration on very clean surfaces. Most of these studies involve semiconductor surfaces or adatoms on semiconductor surfaces. A few studies were concerned with clean copper surfaces (8, 9), and only one paper was listed in which systems with Ni-adatoms on polycrystalline W- and Mo-surfaces were studied.
As stated above, one of the major results of the study of electromigration is, that the movement of atoms is strongly influenced by the momentum transfer from electrons to ions. Only non-elastic collision can produce such a momentum transfers. Systems, in which collisions between electrons and ions are mostly elastic, should show low electromigration. This would be found in systems, in which the surface resistance is very low.
It seems that this concept of low surface resistance as failure inhibitor has not been considered. This may be due to the fact, that the scattering process of electrons at the surface is difficult to modify, so it did not seem to be an adjustable parameter.
The surface resistance in metals can be studied in resistance measurements of thin films. Films less than 2 nm thick are usually discontinuous, 2 to 6 nm thick films are semicontinuous (10). The resistivity of thin continuous films is in first approximation inversely proportional to the thickness t. For instance, Fuad et al. (11) found for chromium prepared at base pressures in the 10
−6
Torr range:
p=
2·10
−8
/t
(&OHgr;
m
2
)
with t the thickness of the film measured in meter. If t≧80 nm, the resistivity is usually constant. It is larger than the bulk resistivity, because the thin film has a high defect density.
Surface scattering can be modified, if one adds overlayers on a thin film. This leads usually to a resistance increase. For instance, if one adds to an annealed Au-film a fractional monolayer of gold atoms (12), the resistance R of this film increases, in spite of the fact that the average film thickness increases. The resistance increase can be explained with an increase in non-elastic electron scattering. The annealed Au-film has in most sections rather flat surfaces. Add-atoms, even if they are from the same material, destroy this smoothness, and produce an increase in the resistivity. If one adds an overlayer of different atoms to a film, the resistance of the film will usually increase, even if the overlayer atoms are metallic. This resistance increase would not be expected in simple theories, where
TABLE 1
Thickness of
&Dgr;R/ due to a 0.7 nm
Substrate
Substrate
Overlayer
thick overlayer
Silver
15
nm
Ge
+20%
Gold
22.3
nm
Permalloy
+12%
Aluminum
10
nm
Germanium
0%
Copper
12.9
nm
SiO
2
0%
Copper
12.9
nm
Chromium
−4%
one would assume that one has two parallel conductors, each with its given conductance, and the total conductance should be the sum of the conductances of both films. Naturally, the conductance of the overlayer should be very small, since it is very thin, but it should give a positive contribution.
The resistance increase due to overlayers is due to increased non-elastic scattering in the bilayer structure. The scattering of electrons in a sample with a thin overlayer takes place not only on the (“new”) surface, but also on the interface between the original film and the overlayer (the “old” surface).
One exception to the rule that thin overlayers increase the resistance is the case Were Cr-atoms are deposition on Cu-films. Table 1 show that a Cr-overlayer 0.7 nm thick reduces the resistance of the film by 4%. This is about the value expected from the film thickness change.
We measured the influence of a chromium overlayer of 0.3 nm on a 4.3 nm thick copper film and found a strong resistance decrease. The relative resistance change was −46%. The resistance decreased in a 6 &mgr;m thick Cu-films by 60% if a 1 &mgr;m thick Cr-overlayer was added. These changes are much stronger than the 4% decrease found by Chopra and Randlett (13). Chopra and Randlett's result can be explained with the thickness increase in their sample. Our resistance decrease is much stronger and represent a new discovery.
That chromium overlayers can influence electromigration was found in the case of silver conductors by Rosenberg et al. (14). These authors studied grain boundary contributions to transport. They studied damage mechanisms, effects of diffusivity, vacancies and grain boundaries on electromigration. They described one test, in which they covered the top of an Ag-film with a 10 nm thick diffusion resistant chromium overlayer to avoid grooving on the silver film. The side of the Ag-film was bare. After subjecting the film to a high current, the top surface covered with chromium showed no defects due to electromigration, whereas the bare edge of the silver film showed a growth pattern. However, it is possible that the edge of an uncoated film shows more defect growth than an (uncoated) top surface because the sharp curvature of the edge surface changes the surface energy (higher vapor pressure, different vacancy concentration than other sections of the film). The typical studies for failure, namely the deteriation of the silver film with time as indicated by resistance changes was not studied, nor was the time to failure measured. No suggestion was made that chromium was superior to other (metallic) overlayers to reduce electromigration defects.
We studied magnetic properties, the Hall effect, the Seebeck effect, and the electrical resistivities of Cr-films on different substrates and with different overlayers over many years (15-23). Some magnetic properties were unusual (15-18). The magnetic diamagnetism was much larger than found in any other materials, except superconductors (15-18). This indicates that the films are most likely partly superconducting. The Seebeck coefficient (19) and the Hall coefficient (20) were close to zero for very thin Cr-films on Ge-substrates, charac
Nelms David
Nhu David
LandOfFree
Chromium films and chromium film overlayers does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Chromium films and chromium film overlayers, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Chromium films and chromium film overlayers will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2938796