If the pedestal is isolated, it may gain a charge as it is hit with charged particles under vacuum. When you open the chamber the charge may dissipate through the air so you may not be able to ever measure any potential when the chamber is open. You need to have a way to dissipate the charge while under vacuum.

Hi Mikhail,

To answer your question, i have one question: do you use RF, DC or DC pulsed magnetron?

Hello Peter Knapen,

DC, but I also tried RF. However, I cannot say anything about the change in the deposition rate from the RF-gun, I do not have the previous data.

The whole story:

First, I used the following parameters : O2 flow rate= 5 sccm, Ar= 20 sccm, P= 3.10-3 Torr, P=200W, tdep= 10 min(for the thickness of ~ 50 nm; in the labmate's work the rate was 5-30 nm/min depending on the oxygen flow rate[however he used 3 Ti targets]. The temperature was 300 C (I wanted to obtain amorphous films). However, the thicknesses were

Hi Mikhail,

First of all, you cannot reactively sputter reactively TiO2 from a Ti target with DC magnetron since the TiO2 will form on the target and block any current going through the target. This can only be done with DC-pulsed generators or RF. Normally if you sputter Ti metal with magnetron sources you should get a deposition rate ~6-10 nm/min, which is close to what you find with two targets and low power (200W which is extremely low for DC-magnetron, usually these magnetrons run at 2-3 Kw for metals). If you use an RF generator the dep rate drops by a factor of 10, so close to 0.7-1.0 nm/min.  If you do reactive sputtering this is always slower than non-reactive sputtering. Sputttering from an oxidic target should also give you deprates in the order of 01-0.7 nm/min depending on RF-power. So in conclusion, the rates you mention are not so different from what is reported in literature for DC and RF dep rates. That your rate dropped from 30 nm/min to 0.5 nm/min in the DC magnetron is because your target is getting oxidized and you block current going through the system. Eventually the DC-plasma will stop. To do reactive sputtering of TiO2 from Ti target you should change to the slower rate RF plasma or put a pulsed DC generator on the magnetron system.

Dear Peter Knapen,

The target poisoning was the first thing we checked. We tried different oxygen flow rates. So, before it was like in the fig. attached. Now all the values are 10 times less.

Moreover, we have been depositing Tio2 layers by reactive magnetron sputtering for so long and the rate has never been that low.

However, we have fixed the system recently and it may be concerned with that. Just do not know where start looking for the answer. Directions of the guns relative to the pedestal? The Z-coordinate of the pedestal? The pollution of the chamber? Preliminary, the shape of the plasma channel has also changed. Now it is located between the shutter and the target.

Hi Mikhail,

If you always have been capable of doing this and it suddenly dropped and you mention the plasma channel has changed I think it may be an electrical problem. Maybe you have a parasitic plasma because a dark space shield is not correctly placed or some ceramic isolators have been sputtered and become conductive. That means that most of your power is lost through a short circuit. Did you notice a drop in the operating voltage by any chance? I would check if all components are still in working order (isolators, shields) and you can check that easily with an ohm meter. Sometimes things get electrical shorted after doing many depositions. The only way to get rid of this then is to clean the components in the system.  

Dear Peter Knapen,

We check the components before each experiment with an ohm meter, it is fine! The chamber and the shields have the same potential, the target- different.

As for the working voltage, I cannot say at the moment. All I may say is that the working power has not changed.

The power will probably be ok, the generator will put out the power, but a different voltage may be an indication of an electrical leak (assuming the flow/pressure is the same as always). Since you mentioned the plasma channel is different this may be an indication of a (parasitic leak).

Thanks, Peter Knapen!

We will start with that then!

May it also be concerned with the magnets used? We noticed that the deposition rate's deterioration is bigger with temperature.

Hi Mikhail,

Yes, this can have a large effect. Some magnets are very temperature sensitive. Usually the magnets are in the cooling water circuit. I have had the experience that if you once do not cool enough (no water running!) you fry the magnets and the dep rates drops. AlNiCo magnets can handle quite a bit of temperature but the strong NiFeN and SmCo magnets loose there strength at low temperatures. Its always good to know what the field strength is right at the target surface. I used to measure this with a Gauss meter to make sure the magnets were still ok.

It would explain the different plasma configuration you mentioned since the confinement of the plasma on the target will change (disappear) if the field drops. But this also should be accompanied with in an increase of the target voltage (at same power level).

Hi Mikhail,

The amount of O2 needed to get a stoichiometric TiO2 depends on the efficiency of dissociation rate of the O2 gas. too much O2 and the deposition rate will drop as you know. So this got me to think if the O2 to Ar ratio changed after you did your system update? reasons for this could be for example that the pumping speed has changed? It looks like your problem is hardware related...

Good luck

Hi Mikhail, I agree with Mr. Alami. Also a change in substrate (polymeric substrate containing water, glass substrate with more adsorbed water, ...) and its water content - as a source of oxygen - can have a drastic effect on where you are on the hysteresis-curve of your reactive process. Therefore, the deposition rate will drop.

We are able to obtain TiO2 in another system of ours. That system has 3 guns and the rates were not lower when we used the same targets from the system with the issue(5 targets system).

Moreover, There are no problems with magnets after all. I checked magnets in the 5 targets system and in 3 targets system. It seems like the magnetic field in the latter is lower than in the former.

What else should we check?

In previous works(~1 year ago), the deposition rate from

3 Ti targets with 0 sccm of O2 introduced was 15 mm/ min. In mine

recent experiments from 2 targets it was 10 mm/min. Seems reasonable.

However, the deposition rate with introducing O2 increased to 30 nm/min

at 5 sccm and then dropped(poisoning effect)[see attached graph]. Now it

drops suddenly at 4 sscm to 0.1 nm/min [though we did not check less

sccms of O2].

Hi Mikhail,

You always have some hysteresis in the poisoning graph. If you start with a clean Ti target (by sputtering with Ar) and then add O2 you will probably find the graph you showed. If you come done from 7 sccm to 0 sccm O2, I think the deprate will go up but probably somewhat below 5 sccm (were you see the drop now). The 0.1 nm/min is indeed low compared to the ~1 nm/min for poisoned target. I would go back to 6-7 sccm and see if the deprate returns to 1 nm/min. If it stays at 0.1 nm/min you can have 2 problems;

1. You have some electrical (parasitic) leak in your system and not all the energy is dissipated in the plasma.

2. More likely and I mentioned that before, you are sputtering with DC-bias a non conductive/low conductive TiO2. TiO2 is a bit strange in the sense that depending on the "conduction" of your TiO2 layer, and that is very much dependent on your poisoning you could create either conductive TiO2-d or stoichiometric TiO2 on the poisoned target. In the latter case you will not get any power in the plasma anymore (it is a real insulating layer TiO2 and DC won't pass) and this might explain the very low deprates. The only solution to this problem is using RF or DC-pulsed generators. Then the stoichiometry of the TiO2 doesn't matter.

Hello Mikhail,

Have you figured out why at last? Been having similar problem but I am reactively sputtering TiN with Ti in Ar/N2 atmosphere. Thank you!

In order to work in the intermediate regime (and have acceptable deposition rate) between metallic mode and poisoned mode, you need some kind of stabilization control: using either plasma emission monitoring on a strong Ti emission line or voltage control with a lambda-sensor.

Paul Lippens Thank you for your suggestion. I’m not familiar with lambda sensor, but I do see voltage on the target. I read somewhere that one way I could try is by having the plasma on and shutter open, at constant power and Ar flow, incrementally increase N2 flow to monitor sudden sharp changes in the voltage to roughly identify each regime. Is that what you are suggesting here? Thank you!

For any reactive deposition, you first need to collect at least a 'voltage curve': start in metallic mode with pure Ar. Then add gradually the reactive gas in steps: record the discharge voltage. This will learn you where (for which flow) the transition to the poisoned mode happens. The deposition then needs to be done (but clean your target first in Ar again) for that reactive flow where you are just on the shoulder of the voltage curve. Problem is that if you don't have a negative feed-back control system, any instability in your process (e.g. a sputter arc) will send you to the poisoned mode. So probably, you will be able to sputter a few minutes, but after that you will end unwillingly in the poisoned mode (with low dep rate).

Paul Lippens Thank you for your suggestion! Could you pls elaborate on this negative feedback control? How would I be able to tell whether I have it? Thank you!

This is an electronic system which measures a plasma parameter (for instance intensity of an optical emission line) and uses that information to control the reactive gas flow in a fast way (PEM-system)