WO2000029636A9

WO2000029636A9 - High purity tantalum targets for sputtering

Info

Publication number: WO2000029636A9
Application number: PCT/US1999/026290
Authority: WO
Inventors: Binxi Sun; Tony Chiang; Vikram Pavate; Peijun Ding; Barry Chin; Arvind Sundarrajan; Ilyoung Richard Hong
Original assignee: Applied Materials Inc
Priority date: 1998-11-12
Filing date: 1999-11-05
Publication date: 2001-07-19
Also published as: JP2002530526A; WO2000029636A2; GB2359825A; WO2000029636A3; KR20010080437A; TW520401B; GB0112058D0; DE19983727T1

Abstract

Improved tantalum-containing barrier layers are obtained by sputter depositing tantalum and/or tantalum nitride from a target having a low contaminant level, i.e., a metallic contaminant content below about 30 ppm by weight; or a niobium contaminant content below about 50 ppm by weight, preferably below 10 ppm, a molybdenum content below about 10 ppm by weight, a gold contaminants below about 15 ppm by weight, and a tungsten content below about 10 ppm by weight. Medium time to failure of copper capacitors including these layers is increased over the use of conventional tantalum targets which have higher amounts of contaminants.

Description

HIGH PURITYTANTALUM TARGETS FOR SPUTTERING

This invention relates to the deposition of improved barrier layers for copper metal lines and contacts for the manufacture of semiconductor devices. More particularly, this invention relates to depositing tantalum-containing barrier layers having enhanced barrier performance.

BACKGROUND OF THE INVENTION In the manufacture of semiconductor devices, conductive metal contacts and lines are deposited over dielectric layers, such as silicon oxide. Heretofore aluminum has been the metal conductor of choice. Since aluminum diffuses into silicon during elevated temperature processing, a barrier layer, particularly one including titanium nitride, is conventionally deposited between the substrate and the aluminum to prevent diffusion or "spiking" by the aluminum into the substrate.

Copper is a better conductor than aluminum, and it has a higher resistance to electromigration than aluminum. However, copper reacts with silicon and copper diffuses into various dielectrics, such as silicon dioxide, at elevated temperatures and under an applied electric field. Thus a good barrier layer is as essential for copper lines and contacts as when aluminum is used as the conductor. Tantalum has been tried as a barrier layer for copper. It is a good conductor and a good wetting agent for overlying copper layers, and it is also a very good barrier to prevent the diffusion of copper into the substrate. Tantalum nitride, formed by sputter depositing tantalum m the presence of nitrogen gas, is a better barrier than tantalum, but it has a higher resistivity than tantalum. Since both tantalum and tantalum nitride have a much higher resistivity than copper, their use detracts somewhat from the advantages of using copper as the conductor. In order to take advantage of the excellent conductivity of copper, a tantalum or tantalum nitride barrier layer must be confor al and as thin as possible .

Conventional sputtering, particularly into small diameter, high aspect ratio openings, has been found to be inadequate to deposit thin conformal coatings into such openings. Conventional sputtering is carried out in a high vacuum chamber using a target of the material to be .sputtered, which is connected to a source of DC power. A substrate is mounted on a support that is spaced from and parallel to the target and argon is passed into the chamber. Permanent magnets affixed to the backside of the target attract argon ions to the target surface after it is powered, where these argon ions impact and sputter off particles of the target material. These sputtered particles then deposit on the substrate. Unfortunately however, sputtering does not occur only in the vertical direction, but in all directions except the horizontal. Thus when high aspect ratio, small diameter openings are to be filled, fewer sputtered particles deposit on the bottom and sidewalls of the openings than on the top of the openings. This is illustrated in Fig. 1 which illustrates the buildup of target material 10 on the top 12 and upper sidewalls 14 of a high aspect ratio opening 16. This buildup prevents many sputtered particles from reaching the bottom 18 and the bottom sidewalls 20 of the opening 16. The resultant coating, as of a barrier material, is not as conformal as is required.

In order to improve the verticality of sputtered particles, an improved sputtering chamber has been developed. A high density plasma is formed in a sputtering chamber between the target and the substrate by means of an inductive coil coupled to a source of RF power. As particles are sputtered from the target, they pass through a plasma region in the vicinity of the coil and become ionized in this region. When the substrate is biased, as by powering the substrate support, the substrate becomes negatively charged; the positively charged sputtered ions formed in the plasma region are attracted to the substrate and they impact the substrate in a more perpendicular direction. Thus more of the sputtered particles deposit on the bottom and bottom sidewalls of openings in this type of chamber, significantly enhancing the bottom coverage, and leading to more conformal sputtered layers . This improved sputtering chamber is known as an "ionized metal plasma" or "IMP" chamber, as shown in Fig. 2. This IMP chamber 170 includes a conventional target 172, as of tantalum, mounted on a top wall 173 of the chamber 170. A pair of opposing magnets 176, 178 are mounted over the top of the target 172. A substrate support 174, bearing a substrate 175 thereon, is mounted opposite to the target 172. A source of power 180 is connected to the target 172 and a source of RF power 182 is connected to the substrate support 174. A controller 200 regulates gas flows. A helical coil 186, which can have one or more turns, preferably made from the same material as the target 172, is mounted between the target 172 and the substrate support 174, and is also connected to a source of RF power 188. Gases such as argon and nitrogen in vessels 192, 194, are metered to the chamber 170 by means of gas flow valves 196, 198 respectively. The pressure in the chamber is maintained by a cryogenic pump 190 through inlet 191 via a three-position gate valve 199.

Providing that the pressure in the chamber is fairly high, i.e., about 10 to a few hundred millitorr, the internal inductively coupled coil 186 provides a high density plasma in the region between the target 172 and the support electrode 174. If the pressure is too low, too few particles are present and sufficient metal lonization will not occur in the region of the powered coil. The gate valve 199 is used to regulate the pumping speed and in turn regulate the pressure m the chamber 170 to the desired range, generally about 10-100 millitorr.

Although the use of the IMP chamber has resulted m the deposition of more robust barrier layers, and better conformality of barrier layers, various problems remain. Improvements continue to be sought in the deposition apparatus and method of deposition to improve tantalum-containing layers as barriers for copper lines and contacts. SUMMARY OF THE INVENTION

We have found that the impurity level in a tantalum target has a large effect on tantalum-containing barrier layer properties. By using a tantalum target of high purity in an IMP chamber, the lifetime and reproducibility of MOS capacitors made with tantalum-containing barrier layers for copper lines and contacts are significantly improved. Improvements are also noted when the coil of an IMP chamber is also made of a high purity tantalum material . BRIEF DESCRIPTION OF THE DRAWING Fig. 1 is a cross sectional view of an opening partially filled with material in accordance with prior art processes. Fig. 2 is a schematic cross sectional view of a modified physical vapor deposition chamber useful in the present invention.

Fig. 3 is a schematic cross sectional view of a test MOS capacitor .

Fig. 4 is a graph of cumulative probability versus time to failure using a prior art target.

Fig. 5 is a graph of cumulative probability versus time to failure using an improved target of the invention.

Fig. 6 is a graph of cumulative probability versus time to failure of three tests using a prior art target. Fig. 7 is a graph of cumulative probability versus time to failure of two tests using the improved target of the invention.

DETAILED DESCRIPTION OF THE INVENTION The robustness of a tantalum-containing barrier layer for copper lines can best be characterized quantitatively by electrical testing of test MOS capacitor structures having a structure Si/SiOx/TaN bamer/Cu using a bias temperatures stress test (BTS) at 275°C and 2MV/cm. A suitable test metal- oxide semiconductor (MOS) capacitor is shown in Fig. 3. A silicon substrate 110 is covered with a dielectric layer 112, such as silicon oxide, generally about 1000 angstroms thick. A thin barrier layer 114 is deposited and a conductive copper metal layer 116 is deposited over the barrier layer. A cap layer 118 is deposited over the copper layer 116 to which a voltage can be applied. The current through the oxide layer is measured. When copper diffuses through the barrier layer into the oxide layer, a catastrophic failure of the oxide is noted. The median time to failure (MTTF) is used to evaluate the barrier performance. The failure distribution gives the standard deviation of the data points.

We have found that by using ultra-pure tantalum as the material used for the target and/or the target and coil in an IMP chamber, improved tantalum barrier layers are obtained. To illustrate this discovery, the impurity level of both metallic and non-metallic contaminants for several tantalum targets were measured. The impurity level of each of the contaminants and their concentration are listed in parts per million by weight (100 ppm = 0.01 wt%) The results are set forth in Table I below. TABLE I

Impurity Target 1 Target 2 Target 3* Target 4

Al 10 5 <0.1

Co 20 5 <0.1

Cr 20 5 <0.1

Cu 20 5 <0.1

Fe 50 15 20 <0.1

K 3 0.1 <0.1

Mn 5 5 <0.1

Na 5 0.001 <0.1

Ni 50 5 10 <0.1

Sn 5

U 0.001 <0.5

N 50 15 10 10

• 200 15 50 15

S 40 <0.1

Zr 5 <0.1

W 25 30 2.5

Ti 10 5 10

Si 25 15 <0.1

Nb 150 30 8.1

Mo 100 40 <0.1

Mg 10 0.5 <0.1

Ca 10 5 <0.1

B 2 <0.1

Li 1 0.001 <0.1

C 50 10 50 10

H 5 10 3

Au <10

Th <0.5

Total Metallic Impurities Reported

339 295.6 100 216

Total Non-Metallic Impurities Reported

340 47 120 38

Total Impurities Reported 679 342.603 220

* Incomplete Analysis

Testing using test capacitors made from the above targets showed that barrier layers obtained using Target 1 gave poor results. The performance of Target 2 was acceptable. Target 3 gave improved results over capacitors made using Targets 1 and 2 ; this target had less niobium present although more oxygen and carbon were present. Thus apparently the metallic impurities have a greater adverse effect on barrier layer quality. Target 4, the purest target material, gave the best barrier results.

Target 2 was used to prepare copper capacitors by depositing 200 angstroms thick films of TaN as the barrier layer. Fig. 4 is a graph of the cumulative probability versus time to failure in hours for capacitors made using target 2. The median time to failure was 6.0 hours.

Fig. 5 is a graph of the cumulative probability versus time to failure in hours for capacitors made using Target 4. The median time to failure was 8.0 hours.

The above tests were repeated, three times using Target 2 and twice for Target 4. The results are shown in Figs. 6 and 7 respectively. A comparison shows that the repeatability of the results using target 4 (Fig. 7) is much better than the results for Target 2.

Although the reason for the improved results using a tantalum target having fewer impurities is not known with certainty, it is believed that intermetallic compounds of tantalum and niobium and/or other metal contaminants, such as tungsten, titanium and aluminum, as well as compounds of metal contaminants and certain atmospheric contaminants, form during deposition that adversely affect devices made with these tantalum nitride barrier layers. Thus in order to form improved targets useful for depositing tantalum nitride barrier layers for copper that have good lifetime and good repeatability, the total amount of metallic impurities present, as well as the type of impurity present, must be carefully regulated. In general, in terms of parts by weight, less than 50 ppm of niobium, tungsten, molybdenum and other metallic impurities should be present for example, and preferably less than 10 ppm. However, the choice of target material is performance based, and the presence of contaminants in a tantalum target that do not form intermetallic materials that are poor barriers for copper probably will not adversely affect the performance of devices made using such a target . Thus the amount of metal contaminants in the tantalum is to be kept below a total of 300 ppm, and preferably below 100 ppm. Less than 15 ppm of gold, less than 10 ppm of tungsten, less than 10 ppm of molybdenum, and preferably less than 10 ppm for all other metal contaminants should be maintained in the tantalum target. Oxygen levels should be kept below 100 ppm and nitrogen levels should be kept below 100 ppm. Although specific data are given above for target materials, the same low contaminant -containing tantalum material should be used to make the coil and pin in an IMP chamber to ensure that the deposited TaN films will have good barrier performance characteristics.

Thus ensuring a low contaminant level in the tantalum material used to make the target and the coil in an IMP chamber ensures that barrier layers made from such tantalum targets are robust, effective barriers for copper metal lines and contacts.

Although the target materials of the invention have been described with some particularity, as explained above, the criteria for particular contaminants are unknown, and are performance based. Not all non-metallic contaminants have the same deleterious effect on barrier properties as metallic contaminants, and some metal contaminants, or their compounds with non-metallic contaminants, may be more injurious to barrier properties than other metals. It is believed that a maximum of 350 ppm of non-metallic contaminants such as carbon, nitrogen, oxygen, hydrogen and the like, should be maintained. Thus one skilled in the art can readily determine whether a particular tantalum material is useful in accordance with the invention.

Claims

1. A tantalum target for a sputtering chamber useful for making tantalum-containing barrier layers wherein said tantalum includes only minor amounts of metallic contaminants.

2. A tantalum target according to claim 1 wherein said tantalum includes no more than 300 ppm by weight of metallic contaminants .

3. A tantalum target according to claim 2 wherein said tantalum includes no more than 100 ppm by weight of metallic contaminants .

4. A tantalum target according to claim 1 wherein said tantalum includes no more than 50 ppm by weight of niobium.

5. A tantalum target according to claim 4 wherein said tantalum includes no more than 10 ppm by weight of niobium.

6. A tantalum target according to claim 4 wherein said tantalum target includes no more than 10 ppm by weight of molybdenum, no more than 10 ppm by weight of tungsten, no more than 15 ppm by weight of gold and less than 10 ppm by weight for other metal contaminants.

7. A tantalum target according to claim 1 wherein said tantalum target includes no more than 350 ppm by weight of non-metallic contaminants.

8. A plasma sputtering chamber including a tantalum target that includes no more than 300 ppm by weight of metallic contaminants .

9. An ionized metal plasma sputtering chamber including a tantalum target that includes no more than 10 ppm by weight of niobium, no more than 10 ppm by weight of molybdenum, no more than 10 ppm by weight of tungsten, no more than 15 ppm by weight of gold and less than 10 ppm by weight for other metal contaminants .

10. An ionized metal plasma sputtering chamber according to claim 9 including a tantalum target that includes no more than 200 ppm by weight of metallic contaminants, a substrate support parallel to and opposed to said target and a coil mounted therebetween wherein the coil is made of the same tantalum material as the target .

11. An ionized metal plasma sputtering chamber according to claim 10 wherein said tantalum target includes no more than 10 ppm by weight of niobium, no more than 10 ppm by weight of molybdenum, no more than 10 ppm by weight of tungsten, no more than 15 ppm by weight of gold and less than 10 ppm by weight for other metal contaminants.

12. An ionized metal plasma sputtering chamber according to claim 9 wherein said tantalum target includes no more than 350 ppm by weight of non-metallic contaminants.

13. A capacitor including a tantalum-containing barrier layer deposited by sputtering a tantalum target including no more than 300 ppm by weight of metallic contaminants.

14. A capacitor according to claim 13 wherein said tantalum target includes no more than 10 ppm by weight of niobium, no more than 10 ppm by weight of molybdenum, no more than 10 ppm by weight of tungsten, no more than 15 ppm by weight of gold and less than 10 ppm by weight of any other metal contaminant.

15 A capacitor according to claim 13 wherein said tantalum target include no more than 350 ppm by weight of non-metallic contaminants .

16. A method of improving the diffusion barrier performance of tantalum-containing barrier layers for copper comprising sputter depositing said tantalum-containing barrier layer by sputtering from a target that includes no more than 300 ppm by weight of metallic contaminants.

17. A method according to claim 16 wherein said target include no more than 50 ppm of niobium.

18. A method according to claim 16 wherein said target includes no more than 10 ppm by weight of niobium, no more than 10 ppm by weight of molybdenum, no more than 10 ppm by weight of tungsten, no more than 15 ppm by weight of gold and no more than 10 ppm by weight for any other metal contaminant.

19. A method according to claim 16 wherein said tantalum- containing barrier layer includes a tantalum nitride layer.

20. A method according to claim 16 wherein said target includes no more than 350 ppm by weight of non-metal contaminants .