WO2001073153A1

WO2001073153A1 - Method of depositing metal films

Info

Publication number: WO2001073153A1
Application number: PCT/GB2001/001407
Authority: WO
Inventors: Paul Rich; Mark Graeme Martin Harris
Original assignee: Trikon Holdings Limited
Priority date: 2000-03-29
Filing date: 2001-03-27
Publication date: 2001-10-04
Also published as: AU4262201A

Abstract

The invention relates to methods of sputter deposition of a metallic layer on to a substrate including utilising substantially krypton.

Description

Method of Depositing Metal Films Sputtering is a well-known process whereby an inert gas is used to sputter a target material onto a substrate. The target is biased with either RF or DC voltage and ions of the inert gas impact the target under the affect of the potential difference on the target, ^sputtering' target material away. There are a relatively few inert gases to chose amongst and argon is in almost exclusive use due to its low price and wide availability. Other noble gases have been tried for sputtering aluminium but no benefit has been observed and therefore for reason of costs and availability they have not been adopted.

Typically noble metals for forming electrical contacts have been evaporated due to their high price. Evaporation processes use relatively small sources of material and recovery of wasted material from the process chamber is relatively easy using aluminium foil liners. Further, no gas is consumed in the process, unlike sputtering that requires a working gas. However sputtering is a preferred process as it produces higher quality films, and is easier to automate. It is a process which is well understood by users. Unfortunately attempts to sputter gold at low temperatures (i.e. less than 120° C) with argon have resulted in films which have unacceptable poor film characteristics e.g. their resistivity is above 3 micro-ohm cm. As a result sputtering of noble metals is relatively unusual and comparatively little researched. The alternative, proposed in the prior art is to sputter with xenon, a gas that costs approximately 25,000 times more than sputter grade argon. In physical vapour deposition processes e.g. sputtering it is widely known that increasing the substrate temperature improves film quality and as a result almost all sputtering processes are onto substrates heated to typically over 150°C and frequently much higher. These high temperatures in general present no problem to the substrate. However there is a process procedure known as ^Λlift-off where a resist pattern (typically organic) is created and then the chosen material deposited onto the thus patterned substrate surface. After deposition the resist is ^Λlifted off by e.g. use of solvent, removing not only the resist but also the material deposited upon it thus leaving a pattern of deposited material. This technique avoids the use of an etchant for the deposited material and is a method of choice for noble metals for obvious reasons however temperatures of over 100-120°C damage the resist and therefore this represents a maximum substrate temperature severely limiting standard substrate heating procedures for improving the electrical properties of deposited metals. The prior art relating to the choice of sputter gas consists in three parts. Firstly that there are several possible gasses for use in sputtering, being the noble gasses. Secondly that RF sputtered gold has an improved quality when sputtered with xenon. And thirdly that in a reactive sputtering process (where, necessarily another gas other than the inert sputter gas is present) the grain structure or adhesion of the resultant metal nitride sputtered film is improved by using specified mixes of the sputtering gasses. Summary of the Invention

Thus from one aspect of the invention is a method of sputter depositing conductive metallic layers on to a substrate utilising an inert process gas characterised in that the process gas is predominately krypton.

As is set out below experiments have now been carried out with a range of non-radioactive noble gasses which indicate that as the mass of sputter gas is increased, there is a significant decrease in the resistivity of the deposited film. It can however also be seen in figure 1 reproduced from page 84 of Glow Discharge Processes, Brian Chapman 1980, ISBN 0-471-07828-X that as the mass of the sputter gas increases the secondary electron yields, the primary cause of ionisation, reduces. This reduction in ionisation efficiency creates practical limits on equipment as higher voltages are required to achieve acceptable sputter rates, or the sputter gas pressure must be increased. As chamber pressures are increased so there is increased inclusion of sputter gas in the deposited film leading to reduced density and increased resistivity. From another aspect the invention consists in a method of depositing a noble metal on to a substrate comprising sputtering of the metal utilising a process gas characterised in that the process gas is at least predominately krypton.

Description of embodiments

To illustrate the effect of substrate temperature with noble metal deposition a gold sputtering process using argon has been run onto semiconductor wafers. The gold, when deposited for 1 minute on wafers at low temperatures (platen temperature of 50°C) , was of 4.5 micro ohm cm. resistivity and was of 2.5 micro ohm cm. resistivity when deposited for 1 minute on warm wafers (a platen temperature of 150°C) all other process conditions identical.

Thus where low substrate temperatures are a requirement a novel process producing a high electrical quality metal layer (e.g. low resistivity) is needed. These low temperatures may be required where the substrates are coated with organic layers damaged by temperatures typical of metal deposition processes. These organic layers could be for example lift-off resist patterns or organic containing low-k dielectric materials e.g. carbon or methyl doped silicon dioxide. Krypton and Argon were experimented with in a Gold deposition process giving the following results: Sputter Resistivity Sheet Resistance Resistivity Film Thickness

Gas micro ohm cm. Ohm/sq. non uniformity SEM observed

Argon 3.98 0.265 2.18% 1 sigma l,50θA Krypton 2.71 0.164 3.35% 1 sigma l,65θA

Process conditions as follows: Process time 1 minute

Target power 2k DC Process pressure 13.5 millitorr

Platen temperature -16°C In another experiment:

Sputter Resistivity Sheet Resistance Resistivity Film Thickness Gas micro ohm cm. Ohm/sq. non uniformity SEM observed

Argon 4.53 0.197 2.22% 1 sigma 2,30θA

Krypton 2.50 0.100 2.60% 1 sigma 2,50θA

Process conditions as follows: Process time 3 minutes Target Power lk DC Process pressure 3.5 millitorr Platen temperature -16°C

In these experiments the substrates were not clamped to the platen and as a result are at a higher temperature rising through time to an equilibrium. During the process the semiconductor substrate did not exceed that which damages lift-off resist. Typical Gold layer thickness used on devices is 2,50θA and a typical process is run at low target powers e.g. 1 k to keep substrate temperatures as low as possible to maintain resist integrity. In addition to the experiments described in the above the following further experiments have been carried out with a range of gasses to sputter gold. The results are as follows: Process conditions:

Silicon wafer substrate lkW target power

2min process time

-16 °C wafer platen temperature Sheet DC Process

Sputter Resistivity Resistance Thickness Target Pressure

Gas μΩcm ΩD A Voltage millitorr

Neon 14.9 1.49 1000 427 3.5 Argon 5.2 0.322 1600 559 3.5 Krypton 2.6 0.155 1700 745 3.5 Xenon 2.8 0.255 1100 785 20

The gasses experimented with are all the non- radioactive noble gasses.

It may be that a key aspect of Gold layer resistivity on a substrate is sputter gas inclusion in the sputtered film during the sputtering process. Krypton may include less than argon due to its larger size. If this explanation is correct then it would be expected that xenon would yield even better electrical quality metal layers for any given applied target power, substrate temperature and pressure. As can be seen, in the same time and with the same applied power krypton yielded the thickest film, and thus the lowest sheet resistance and more importantly the lowest resistivity film (film thickness independent) . Part of the explanation may be found in figure 2, a reproduction of a table appearing in Glow Discharge Processes on page 183. As can be seen there is a monotonic increase in sputter yield with increasing mass of sputter gas, however as figure 1 illustrates, there is also a monotonic decrease in secondary electron emission the primary cause of ionisation to the point where the apparatus was unable to generate sufficiently high DC voltages on the target to cause a glow discharge (plasma) with xenon and thus unable to sputter at all. As can be seen target voltage increased to deliver the selected power at a standard sputtering pressure of 3.5 millitorr as the atomic number of the sputter gas increased. To cause any sputtering with xenon a dramatic increase in chamber pressure (by a factor of approximately 6) was required, thus greatly increasing the quantity of sputter gas in the chamber. It was noted that the resultant film was more resistive and the process slower and electrically less efficient. The increase in resistivity is probably due to greater sputter gas inclusion in the sputtered film. At the time of writing it should also be considered that xenon is 5 times more expensive than krypton, and all other conditions the same 6 times as much gas will be consumed. This renders the process 30 times more costly in gas consumption. By way of contrast argon of the same quality in bulk costs over 5,000 times less than krypton. A typical sputter process might use 0.2 litres of argon per wafer. At the time of writing therefore the sputter gas cost per semiconductor wafer would rise from 0.001 pence per wafer for argon to 53 pence per wafer for krypton and £14.29 per wafer for xenon.

This increase in pressure can also be achieved by reducing the pumping speed. Whilst this avoids the increase in gas consumption, it offers no other benefit and is generally considered undesirable. All vacuum systems leak slightly and the substrates outgas absorbed gasses. Reducing the pumping speed increases the proportion of this gas contamination in the sputter gas thus further reducing sputtered film quality.

It should also be noted that for a given process time and applied power the sputtered thickness was less for xenon than for krypton, at least in part due to increased gas scattering in the higher pressure gas. In effect the target has been sputter eroded, but the sputtered material is wasted on the chamber wall shielding rather than being deposited onto the substrate. This is significant with a target as valuable as e.g. gold or platinum. As can be seen in the experimental results krypton was 54.5% more efficient than xenon at depositing gold onto the substrate and at least a part of this is accounted for by target material eroded but not deposited upon the wafer.

Further, it is noted that aluminium, the most commonly sputtered material for electrical interconnect usage is lighter (at atomic number 13) than argon (at atomic number 18). Aluminium sputtered in argon has a resistivity that approximates to bulk and thus cannot be significantly improved upon by using more massive (and expensive) sputter gasses. In experiments with titanium (atomic number 22) not reported here, it was found that krypton (with an atomic number of 36) gave films of lower resistivity than when sputtered in argon, however the improvement was not as dramatic as for gold. It is therefore hypothesised that for metals with an atomic number metals greater than that of xenon, there is a first order monotonic reduction in sputtered film resistivity as the sputter gas mass is increased. This is countered by ionisation efficiency reductions as the gas mass is increased with a best compromise at krypton. Where the sputtered material is already of lower mass than the sputter gas little improvement is effected, e.g. aluminium and argon. For the metals heavier than argon, but lighter than krypton, an improvement is found by using krypton rather than argon. There can clearly be no reason based on this work to select xenon with its inferior results and considerably increased operating costs. It should be understood that the advantage of using krypton shown here will attach to mixes of krypton and another noble gas and any effective improvement by the inclusion of krypton is to be included in the this invention.

It should also be understood that this invention is particularly advantageous where the substrate, for example a compound semiconductor wafer perhaps with an organic mask in place or any semiconductor wafer with a low thermal budget, cannot be heated to an effective temperature to reduce the sputtered film resistivity when using argon. This invention can thus be seen as a way of reducing process temperature to achieve the same resistivity sputtered films. An electrostatic chuck may advantageously be used to improve platen to substrate thermal transfer enabling more effective temperature control of the wafer. The electrostatic chuck may be chilled to below ambient temperatures.

Claims

1. A method of depositing a metallic layer on to a substrate including a sputter deposition of the layer utilising a process gas characterised in that the process gas is at least predominately krypton.

2. A method of depositing a noble metal on to a substrate comprising sputter deposition of the metal utilising a process gas characterised in that the process gas is at least predominately krypton.

3. A method as claimed in any one of the preceding claims wherein the metal is gold.

4. A method as claimed in any one of the preceding wherein the substrate temperature is maintained .below

120°C 5. A method as claimed in any one of the preceding claims wherein the process time is between 30 seconds and

4 minutes.

6. A method as claimed in any one of the preceding claims wherein the target power is in the range 1 to 6 kW DC.

7. A method as claimed in any one of the preceding claims wherein the process pressure is between 1 and 15 millitorr.