Looks like you are using a SiO2 (oxide target) 

Firstly by cleaning the target with acetone and then wiping it with a clean room wipe, I would say you contaminated your target to a great extent.

Because upon initial sputtering the black grey debris is clearly indicative of Carbon contamination.

Later after extend hours of sputtering of 6 hours, probably the surface contamination got removed. But you have now observed a ring, which is probably dark brownish in color, and the more you sputter in Argon atmoshphere (by any chance), you will find it to be more brownish and dark. Probably sputtering in Ar+O2 will reduce such effects.

In case the t\oxide target is now getting reduced. Which means there is oxygen deficiency in the composition of the target. The deposition rate will be quite low, and if you keep on increasing the power and also the sputtering time, the surface discoloration (indicating) oxygen deficiency will increase.

Best way is to bond a pure silicon wafer (well cleaned and vapor degreased), and sputter it in oxygen rich (Ar+O2) ambient at a convenient RF power. You can probably get some decent films.

K. Sreenivas

Dear K. Sreenivas,

Thank you for your input. I will attempt to clean and sputter it in a O2 rich environment to restore the oxygenation of the target. May I ask, were does this Carbon contamination come from? and how can I prevent it in future experimentation?

Best wishes

David.

Acetone, IPa etc are all organic solvent, even though we think after the target cleaning using such chemical they have evaporated, but there can be some residual carbon.

Secondly if we tough the target surface by our finger or hands, contamination comes.

Thirdly when we wipe the surface with a clean wipe/tissue paper, as the target surface is rough, small fiber can remain sticking to the surface, and these are fibers of organic material.

I wonder if you can fully restore the oxygenation of the oxide target by sputtering in O2 rich environment. Normally it requires annealing, or sintering at high temperatures in furnace.  But then the problem arises, because the oxide target is bonded to a metal blank underneath, and the bonding gets affected.

This is why I have alway resorted to reactive sputtering by sputtering a silicon wafer in an oxygen rich atmosphere.  Even if the silicon wafer cracks, it is inexpensive, I can use another silicon wafer conveniently.

Oxide targets are always quite expensive, and after a few repeated sputtering runs, they become oxygen deficient, and the stoichiometry is affected.

K. Sreenivas

Dear Dr. Coathup,

Provisionally, I disagree with Dr. Sreenivas' reasoning but agree with his final recommendation, which was to reactively sputter from a pure Si target, as one optional, possible solution.

I suggest that your old BTO target material was actually slightly electrically conductive, while the SiO2 is likely a near-perfect insulator.  I would not be surprised if the BTO target had an effective, as-processes bulk resistivity as low as 1 to 10 ohm*cm.  The SiO2 target might have near the bulk theoretical resistivity of over 1E12 ohm*cm.  A small bulk conductivity of an 'insulator' target can even out (or redistribute) local charge.  Of course, surface conductivity can do the same thing.  I think the BTO is more likely to develop surface conductivity than SiO2 would be, as well.  In any/either case, the ability of the target to avoid high local charge build-up may dramatically affect the (+) and (-) charge transport from the plasma to the target.

To review, if the target was perfectly insulating, both bulk and surface, and the plasma contact with the surface was perfectly laterally uniform across the face of the target, and the plasma was quiescent and uniform within itself, then the plasma and the insulative target surface will spontaneously assume a potential difference with respect to each other to (try to) guarantee both quasineutrality of the plasma and zero charge build-up on the surface of the target.    If the plasma is not quasineutral, it means the plasma has unequal volume densities of (+) and (-) charges, and the plasma will 'blow up' (extinguish) because of space charge repulsion of like charges.  If unequal areal densities of (+) and (-) charges are arriving at the target surface per unit time, the surface will charge up.  Since electrons (-) are much more mobile than ions (+), the surface usually charges negatively.  At some (usually low) negative potential, as measured with respect to the plasma potential, sufficient (-) charges are repelled from the negative surface to balance the areal arrival rates of (+) and (-) charges.  This is what you want.  You cannot have anything else, otherwise the potential on the target would go on increasing indefinitely.  Ideally your spontaneously-arising potential difference between the plasma and the target surface would stabilize at a nice few-hundred-volts potential in which the target is negative; then you would sputter nicely while maintaining a net-neutral target surface charge condition uniformly across the face of the target.

However, in a real RF magnetron system, there are magnets embedded at specific places within the target and spatially non-uniform generation of (+) and (-) particles within the plasma.  Also, the plasma is contained by walls or magnetic fields that have real impacts on (+) versus (-) particle losses; the plasma is much different than the infinite, constant-density, uniform 'ideal' plasma filling all space in front of the target, that I reviewed above. 

You see where we're going here?  Local charge transport non-uniformities at the walls (your chimney) and the target will occur.  Local non-uniform and imbalanced charge accumulations on at least the target surface and maybe also the chimney walls (surface) will occur.  All of this was happening with the BTO target as well, but someone way back in pre-history fiddled around with this machine until they struck a balance of conditions that worked for BTO targets.  Or maybe they used exquisite human intelligence and hard-won insight.  People who can 'tame' machines like this for new processes and new materials are highly valuable.

It seems to me, SiO2 is not an exotic material, and the original equipment manufacturer (OEM) for your RF sputtering machine must have provided SiO2 sputtering parameter combinations long ago.  Can you find out what those were?  Or, if the system is a "home brew" machine, there ought to have been someone back in the day who understood how his/her machine worked for a 'perfect' insulator like SiO2.  In fact, they would have used a perfect insulator for their base design calculations.  Can you reach back into the old notebooks and records to find that data?

For a wide variety of reasons, there is no reason to assume that sputtering parameters that worked for BTO would work for SiO2, and vise-versa.  In fact, one would assume the opposite, for very good reasons.  Hopefully you can ask for help and have good SiO2 parameters handed to you.  If not, then you have a nice adventure in front of you.  We will be glad to help.

Hi David,

Very familiar problem. I don't think the staining/colouring of the target has anything to do with the solvent you used to clean it. This will stay a bit on the surface and should disappear as soon as you start the plasma and do a first clean-up run.

1. Most likely you are producing substoichiometric SiO2-d or SiO-like material and that is brown-blackish and will start to conduct after a while. You can perfectly sputter stoichiochiometric SiO2 from a SiO2 target but you should add some O2 to the Ar sputter gas to compensate for the loss of oxygen from the target. Normally I use 1% O2 in Ar and that should suffice to keep the target in good condition. I've sputtered both Al2O3 and SiO2 in the past and if you don't add additional O2 you will not only get coloured targets but you will deposit sub-stoichiometric thin film layers.

2. There is a chance that you sputter some materials from the reactor, shutters or other components which get's on the target (and substrate) but I think this is unlikely. Check if you don't have parisitic plasma's between target and the top of your reactor or between target and (dark space) shields/shutters. If you see area's of flickering plasma's this is a strong indication. The discharge should be located only between the target and substrate table. If this is not the case you might be sputtering from something different than the SiO2 target (and that can pollute the target).

3. The high bias you notice is dependent on a few things: pressure (the higher the pressure the lower the bias), power (the higher the power the higher the bias), the sputter gas (ionisation degree) and target material (conductivity and secondary electron emission). If you have a conductive target the bias will be lower. If you have an oxidic target secondary electrons emission is higher than for metals. So different materials will give different bias when all the other parameters (power, pressure, flow, electrode distance) are kept the same. So the difference in bias you see between targets is related to the material properties.

Normally a high bias will make sure the sputtered species arrive at the substrate with high kinetic energy (which has impact on the morphology of the layer). You can easily drop bias by raising the pressure. Again, this will have impact on the morphology of the deposited layer.

Dr. Coathup,

I like Peter Knapen's answer.  His is much more practical than mine.  I would do what he says, first. 

My answer basically came to, you should ask your colleagues who built the RF magnetron machine, or consult old documentation, to find valid operating parameters for SiO2.

But my asnwer assumed that the RF magnetron source and system were working properly.  Peter's answer allows that something may have gone wrong.  I advise to check for correct mechanical, electrical and plasma conditions, as Peter says.  Beyond that, you might also check the gas distribution into the target/plasma region, e.g., for any disconnected gas tubing.

Even if everything in hardware would check out to be OK, my idea to find an old set of process parameters for SiO2 still has a problem.  Such a "recipe" for SiO2 would undoubtedly start from a known, standard condition for the target and the chimney.  Since you have (maybe) contaminated or degraded these components, they will have to be brought back to their standard state, before the old recipe will work properly.

Both Peter and Dr. Sreenivas recommended adding some O2 gas to the sputtering gas.  I strongly agree.  I always do this, for the reasons they say.

I can confirm here that we also sputter SiO2 from a quartz target in argon with a little oxygen gas added. It should not be very complicated. Maybe you have to play a bit with the amount of oxygen added and check the stoichiometry of the resulting film, preferrably by XPS because it can also give you information on the oxidation state of the silicon atoms in the sputtered film (should be 4+ for stoichiometric SiO2).

Hello everyone,

I am doing RF sputtering of silicon dioxide using a 3 inch SiO2 target in 1% Oxygen in Argon (300W) but despite that I observed a light brownish ring in the target. Other than that the target shield turned black (silvery initially) after 5 runs (~350 nm each). Is it due to the sub-stoichiometric compounds?  I am not exactly sure what is causing it and maybe once I know more I can try to solve it. It would be great if someone could shed some light on it if possible. 

Thanks.

if my substrate has photoresist on it (patterning layer), then adding 1% of O2 will burn the photoresist and damage my pattern? Also I am trying to sputter SiO2 in Ar plasma with Ar flow rate of 10 Sccm that give me a sputtering pressure of 10 ^-2 mbar at the power of 180 W RF, after 30 min sputtering, I got only a few nm of SiO2. I dare not go to high power for fear that my substrate is heated up and the photoresist pattern is damaged. From this paper here, it looks relatively easy: http://iopscience.iop.org/article/10.1088/0960-1317/17/5/029/meta

I did some calculation based on the paper and was expecting at least around 100 nm of SiO2 after 30 min of sputtering with that conditions.

Can someone share some experience, please!

Many thanks

Hi Thuy Thi Thanh Pham,

Are you also using a diode sputtering system? Or is it a magnetron sputtering system? Because that would probably give quite different deposition rates.

For a magnetron sputtering system you have to worry less about your resist being damaged, because the plasma will be confined to the target. If you have a magnetron sputtering system I do not think it would be a problem to go to higher sputtering power or add oxygen.