Patent Application
20060229000
This invention relates to a polishing pad for polishing a target object such as a semiconductor wafer or a magnetic hard disk substrate that requires a high degree of flatness on the surface and more particularly to such a pad suited for the planarization of wafer surfaces in the production process of semiconductor devices.
In the production process for semiconductor devices, the metallic wiring layer for mutually connecting elements such as transistors, capacitors and resistors is coming to be produced in a multi-layer form, and such layers are usually formed by using photolithography or damascene technologies. By the photolithography method, a wiring pattern is exposed to light to produce a multi-layer structure but steps tend to be formed on the surface of the device when an insulating film is placed over a metallic wiring layer, etc. If the surface unevenness of the device due to such steps becomes greater than the depth of focusing at the time of light exposure, the accuracy in the widths and shapes of the pattern is adversely affected and the yield of the semiconductor devices becomes poor as a result. In other words, if the unevenness of the surface unevenness of the device caused by such steps can be reduced, the exposure margin in photolithography can be secured, the minute patterning and etching processes for the wiring layer can be carried out both accurately and easily and hence the yield of semiconductor devices can be improved.
By the damascene method, a wiring metal (Cu) is placed after wiring grooves are formed on the insulating film and a multi-layer structure is obtained by leaving the metal only inside the groove by a polishing process but there remains the so-called problem of “dishing” whereby the center part of the metal is made too thin by the polishing process. For this reason, planarization of the device surface is very important in the production process of semiconductor devices.
Planarization is done by chemical mechanical polishing (CMP). CMP is a polishing technology combining both a chemical operation by using a fabrication liquid and a mechanical removal operation by using abrading particles whereby the device surface is chemically melted by the fabrication liquid and mechanically polished by the abrading particles, and it is widely being used because unwanted parts different from the internal parts are hardly formed on the surface by the process.
Planarization by CMP is carried out by rotating a lapping plate (or a platen) with a polishing pad attached thereto, supplying slurry having small particles of a material selected from the group consisting of silica, alumina, ceria and zirconia dispersed in an alkaline or acidic fabrication liquid on the surface of the polishing pad, and pressing thereto the surface of the wafer (or the surface of the device) attached to a polishing head (or a carrier).
A polishing pad of a foamed material such as polyurethane having many air holes formed by air bubbles at the time of foaming is usually used (such as IC1000 (product name) produced by Rohm and Haas Electronic Materials CMP Holdings Incorporated).
Since foamed pads are easily compressive and deformable, however, they tend to cause remnant steps by becoming deformed and entering indentations on the device surface and abrading the interiors of the indentations during the planarization process. Moreover, since there are fluctuations in the air hole density inside a foamed pad, although the wafer surface is uniformly pressed onto the pad surface, the repulsive force of the pad against this compressive force becomes non-uniform. Thus, the surface of the pad cannot operate on the wafer surface uniformly and the wafer surface cannot be polished uniformly. As a result, fluctuations appear in the stock removal within the polished wafer surface.
For this reason, Japanese Patent Publication Koho 2004-507077, for example, proposed the use of a pad produced by hardening a synthetic resin such as polyurethane without foaming (hereinafter referred to as non-foamed pad). Since non-foamed pads are hard to compress (having a low compressibility) and slow in recovering from a deformed condition under compression (having a small elastic repellent force), they do not easily enter and abrade the interiors of the indentations on the device surface and hence the generation of remnant steps can be reduced. Since the polishing rate is significantly lower with non-foamed pads than with foamed pads, grooves are commonly formed on the pad surface so as to improve the polishing rate.
The polishing rate of a non-foamed pad cannot be improved to be nearly the same as that of a foamed pad, however, merely by forming a pattern of grooves on the pad surface.
Although non-foamed pads were considered capable of uniformly applying their elastic repellent force to the pad surface because there are no air holes formed inside, furthermore, this has turned out not to be the case because, since non-foamed pads have a slow compression recovery rate and are slow in recovering from the compressed condition, there are fluctuations in the elastic repellent force within the surface and a uniform force cannot be applied onto the wafer surface.
It is therefore an object of this invention to provide a polishing pad capable of planarize the surface of a wafer (or the surface of a device) quickly and uniformly.
As described above, it used to be believed regarding polishing pads of a non-foamed material that fluctuations in stock removal increase over the surface of the target object if the compression recovery rate is reduced and that the elastic repellent force increases and remnant steps come to result if the compression recovery rate is made too large. The present inventors have discovered, however, the appearance of remnant steps can be reduced and the target surface can be polished quickly and made uniformly smooth by using a polishing pad of a non-foamed material with a high compression recovery rate and a low compressibility if its hardness is appropriately selected. Thus, a polishing pad according to this invention is characterized as comprising a planar pad main body of a non-foamed synthetic resin, having Shore D hardness of 66.0-78.5, or preferably 70.0-78.5, or more preferably 70.0-78.0, or even more preferably 72.0-76.0, compressibility of 4% or less, or preferably 2% or less, and compression recovery rate of 50% or greater, or preferably 70% or greater.
In order to suppress the occurrence of the hydroplane phenomenon, grooves may be formed on the surface of the pad main body. The surface ratio of the grooves (or the ratio of the percentage of the surface area covered by the grooves) may be 10%-50%.
In order to improve the ability of the pad to follow the surface contour of the target object, a backing sheet having Shore D hardness greater than that of the pad main body and compressibility higher than that of the pad main body may be affixed to the back surface of the pad main body.
The pad main body may be transparent. According to a preferred embodiment of the invention, the pad main body has thickness 0.5 mm-2.0 mm and transmissivity of 10% or greater to light with wavelength of 350 nm-900 nm, and preferably 30% or greater to light with wavelength of 450 mm-900 nm.
With a polishing pad structured as above, a wafer surface can be polished quickly and smoothly without causing fluctuations in stock removal by reducing remnant steps.
Compressibility is the ratio of change in the thickness of the pad main body 11 as the load is changed from 1.4 psi (the standard) to 16 psi under the temperature condition of 23±3° C. According to this invention, compressibility of the pad main body 11 is 4% or less, and preferably 2% or less. If compressibility exceeds 4%, the interiors of the surface indentations of the wafer are abraded excessively, causing remnant steps.
Compression recovery rate is measured as the change in the displacement of the pad main body 11 with a load of 16 psi under the temperature condition of 23±3° C. After the load is reduced to 1.6 psi, the displacement by recovering the thickness in 30 seconds is measured and this measured displacement is divided by the displacement with the load of 16 psi. According to this invention, the compression recovery rate of the pad main body 11 is 50% or over, and preferably 70% or over. If the compression recovery rate is less than 50%, the unevenness on the device surface cannot be followed sufficiently well, and fluctuations result in the elastic repellent force of the pad over the wafer surface such that the wafer surface cannot be polished uniformly.
Grooves may be formed on the surface of the pad main body 11 in order to prevent adsorption of the wafer and to reduce the possibility of occurrence of the so-called hydroplane phenomenon. The ratio of surface area where grooves are formed (by considering the surface of the pad main body 11 to be a flat plane) with respect to the whole area of the pad main body 11 is in the range of 10%-50%. The grooves may be concentric circles or in a lattice formation. They may be any combination of straight and curved lines in any geometrical pattern.
Polishing process using a polishing pad of this invention is explained next. As explained above about the production of a semiconductor device, steps are formed on the surface of a wafer if an insulating film is formed on a metallic wiring formed on the wafer. Such steps are removed in order to planarize the wafer surface according to this invention by using a CMP polishing machine as shown in
As shown in
When the technology shown in
Although not shown, the CMP devices of
The polishing devices 20 shown in
Slurry is obtained by dispersing abrading particles in water or a water-based aqueous solution and further adding a reactive liquid capable of chemically reacting with the wafer surface such as sodium hydroxide and ammonia.
The pressure (or the polishing pressure) with which the wafer 26 is pressed to the surface of the polishing pad 10 (or the surface of the pad main body 11) is within the range of 1 psi-10 psi.
The invention is described next by way of test and comparison examples.
A resin solution was prepared by adding 26.5 parts of 3,3′-dichrol-4,4′-diamino-diphenyl methane as hardener to 100 parts of TDI (trilene diisocyanate) type urethane prepolymer with average molecular weight of about 750. This resin solution was placed inside a mold and hardened inside this mold to obtain a non-foamed block. This block was sliced into a thickness of 1.5 mm to produce pad main bodies with Shore D hardness of 78.0 (at measurement temperature of 23.0° C.). Grooves (of spiral shape with land width 0.6 mm and groove width 0.3 mm and groove occupancy ratio of 33.3%) were formed on the surfaces of these pad main bodies by means of a lathe to produce polishing pads of Test Example 1.
A resin solution was prepared by adding 26.5 parts of 3,3′-dichrol-4,4′-diamino-diphenyl methane as hardener to 100 parts of TDI type urethane prepolymer with average molecular weight of about 900. This resin solution was placed inside a mold and hardened inside this mold to obtain a non-foamed block. This block was sliced into a thickness of 1.5 mm to produce pad main bodies with Shore D hardness of 75.0 (at measurement temperature of 23.0° C.). Grooves (of spiral shape with land width 0.6 mm and groove width 0.3 mm and groove occupancy ratio of 33.3%) were formed on the surfaces of these pad main bodies by means of a lathe to produce polishing pads of Test Example 2.
A resin solution was prepared by adding 26.3 parts of 3,3′-dichrol-4,4′-diamino-diphenyl methane as hardener to 100 parts of TDI type urethane prepolymer with average molecular weight of about 960. This resin solution was placed inside a mold and hardened inside this mold to obtain a non-foamed block. This block was sliced into a thickness of 1.5 mm to produce pad main bodies with Shore D hardness of 72.0 (at measurement temperature of 23.0° C.). Grooves (of spiral shape with land width 0.6 mm and groove width 0.3 mm and groove occupancy ratio of 33.3%) were formed on the surfaces of these pad main bodies by means of a lathe to produce polishing pads of Test Example 3.
A resin solution was prepared by adding 26.0 parts of 3,3′-dichrol-4,4′-diamino-diphenyl methane as hardener to 100 parts of TDI (trilene diisocyanate) type urethane prepolymer with average molecular weight of about 1080. This resin solution was placed inside a mold and hardened inside this mold to obtain a non-foamed block. This block was sliced into a thickness of 1.5 mm to produce pad main bodies with Shore D hardness of 66.0 (at measurement temperature of 23.0° C.). Grooves (of spiral shape with land width 0.6 mm and groove width 0.3 mm and groove occupancy ratio of 33.3%) were formed on the surfaces of these pad main bodies by means of a lathe to produce polishing pads of Test Example 2.
A resin solution was prepared by adding 47.3 parts of 3,3′-dichrol-4,4′-diamino-diphenyl methane as hardener to 100 parts of HDI (hexamethylene diisocyanate) type urethane prepolymer. This resin solution was placed inside a mold and hardened inside this mold to obtain a non-foamed block. This block was sliced into a thickness of 1.5 mm to produce pad main bodies with Shore D hardness of 80.0 (at measurement temperature of 23.0° C.). Grooves (of spiral shape with land width 0.6 mm and groove width 0.3 mm and groove occupancy ratio of 33.3%) were formed on the surfaces of these pad main bodies by means of a lathe to produce polishing pads of Comparison Example 1.
A resin solution was prepared by adding 23.0 parts of 3,3′-dichrol-4,4′-diamino-diphenyl methane as hardener to 100 parts of TDI type urethane prepolymer with average molecular weight of about 1260. This resin solution was placed inside a mold and hardened inside this mold to obtain a non-foamed block. This block was sliced into a thickness of 1.5 mm to produce pad main bodies with Shore D hardness of 60.0 (at measurement temperature of 23.0° C.). Grooves (of spiral shape with land width 0.6 mm and groove width 0.3 mm and groove occupancy ratio of 33.3%) were formed on the surfaces of these pad main bodies by means of a lathe to produce polishing pads of Comparison Example 2.
Commercially available polishing pads of a foamed material (IC100 produced by Rohm and Haas Electronic Materials CMP holdings Incorporated) were used as polishing pads of Comparison Example 3. Their thickness was 1.23 mm and Shore D hardness was 59.0.
Shore D hardness, compression ratio and compression recovery rate of the polishing pads of Test Examples 1-4 and Comparison Examples 1-3 are summarized in Table 1.
Comparison Test 1
Polishing pads of Test Example 3 and Comparison Example 3 were used to polish the surfaces of wafers and the steps on the wafer surfaces were examined for each pad.
Well known test wafers (SKW7-2) commonly used for evaluating planarization were used. These are wafers produced by using a specified mask pattern to etch the surface of silicon substrates and having a silicon oxide film vapor-deposited by a CVD process. In Comparison Test 1, steps of Patterns D20, D40 and D80 were compared with Pattern D100 on test wafers each polished by using a different polishing pad. In the above, Pattern D100 is the portion of a test wafer with no unevenness. Pattern D20 is the portion having straight line protrusions with width 20 μm and straight line indentations with width 80 μm and depth 0.8 μm formed alternately. Pattern D40 is the portion having straight line protrusions with width 40 μm and straight line indentations with width 60 μm and depth 0.8 μm formed alternately. Pattern D80 is the portion having straight line protrusions with width 80 μm and straight line indentations with width 20 μm and depth 0.8 μm formed alternately. Unevenness on the wafer surfaces was measured by using a commercially available apparatus for the measurement of steps (P-1 (product name) produced by Tencor Inc.) For the polishing, a commercially available polishing machine as shown in
The results of Comparison Test 2 are shown in
Comparison Test 2
Polishing pads of Test Examples 1, 2 and 4 and Comparison Examples 2 and 3 were each used to polish a wafer surface as done in Comparison Test 1 (using wafers, slurry and the polishing machine of the same kinds and under the same conditions as shown in Table 2, except that the steps of Pattern D80 were compared with Pattern D100).
The results of Comparison Test 1 are shown in
Comparison Test 3
Polishing pads of Test Examples 1, 2 and 4 and Comparison Examples 1-3 were each used to polish the surface of a wafer with a PTEOS film (with diameter 200 mm) having a plasma oxide film (with thickness 10000 Å) formed on the surface without any pattern. The results were compared regarding stock removal and WIWNU (“WithIn Wafer Non-Uniformity”) for each pad. In the above, WIWNU is a parameter showing fabrication uniformity within the wafer surface and expressed in terms of percentage of (standard deviation in stock removal within the wafer surface)/(average stock removal). A smaller WIWNU value indicates a smaller fluctuation in stock removal.
Film thickness was measured for the measurement of stock removal by using a commercially available light interference type apparatus for measurement of film thickness (Nanospec9200 (product name) produced by Nanometrics, Inc.) The same CMP polishing device and slurry as used in Comparison Tests 1 and 2 were used under conditions shown in Table 3.
The results of Comparison Test 3 are shown in Table 4. Each number shown in
For Comparison Example 3, polishing rate = 2000 Å and WIWNU = 7.2%.
The value which is required to be practical value is 1700 Å/min for polishing rate and 8.0% or less for WIWNU.
As shown in
| Number | Date | Country | Kind |
|---|---|---|---|
| 2004-299838 | Oct 2004 | JP | national |
This application is a continuation of International Application No. PCT/JP2005/012903, filed Jul. 13, 2005 which claims priority on Japanese Patent Application 2004-299838 filed Oct. 14, 2004.
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/JP05/12903 | Jul 2005 | US |
| Child | 11447425 | Jun 2006 | US |
