Hi Mikhail,

It is known that sometimes oxide targets have more difficulty to ignite in RF systems. One thing which usually works is increasing the sputter pressure during ingnition. You can always lower it when the plasma is on. When you use target shutters you can also get into real trouble if the distance between the shutter and target is to small. lowering the shutter a bit might improve ignition a lot. Last but not least, I don't know the Trump RF generator but some generators use a peak ignition voltage to start the plasma. Some systems use a sparking plug to get some ions into the plasma so ignition is enhanced. A pressure burst in the system is usually the best way to get it working.

Once ignition is done on oxide targets it should be much easier to sustain the plasma with oxidic targets then with metals targets since oxidic targets create more secondary electrons which will ionize Ar neutrals again.

Hope this advise helps.

Peter

Hi Mikhail,

It is known that sometimes oxide targets have more difficulty to ignite in RF systems. One thing which usually works is increasing the sputter pressure during ingnition. You can always lower it when the plasma is on. When you use target shutters you can also get into real trouble if the distance between the shutter and target is to small. lowering the shutter a bit might improve ignition a lot. Last but not least, I don't know the Trump RF generator but some generators use a peak ignition voltage to start the plasma. Some systems use a sparking plug to get some ions into the plasma so ignition is enhanced. A pressure burst in the system is usually the best way to get it working.

Once ignition is done on oxide targets it should be much easier to sustain the plasma with oxidic targets then with metals targets since oxidic targets create more secondary electrons which will ionize Ar neutrals again.

Hope this advise helps.

Peter

hello,

When you are using a non-conductive TiO2 target the load (impedance) is changing.

In this case, to make a good matching and plasma to ignite change the length of the coaxial cable (Try to shorten the length) also during matching slowly chnage the gas pressure (try from 3-10 mTorr). Good luck.

Dear Mikhail

 It should not be any problem with ignition of the discharge with oxide target. Moreover, the possibility to sputter the non-conductive target is a one of the main advantage of the RF system. We used this technics  everyday for sputtering of the polymer target. In my opinion, if you can not start the discharge, the problem can be in the electrode ( for example  short circuit) or in the non proper contact between the cables.  You can easily check it with multimeter.

Best regard

Dear Mikhail Pylnev,

As per i understood from your question is you want to get the plasma for oxide materials. 

1. Most of the oxide materials required higher watt. Suppose in my case copper will ignite at around 12 to 14 watt but copper oxide will ignite around 42 watt.

As per i understand most of the oxide are semiconductors in nature hence it required higher power to accelerate the plasma. For your case if your trying your targets in lesser power constantly increase the power at a point you will get plasma. I understand from my RF unit pressure is also playing major role.        

Hope this might be helpful for you.

Enjoy your research!

Hello,

For the ignition, you also need to check the regulation mode. Also, you should put the matching box capacitors at the good position (TiO2) before. Increasing the pressure could be a solution before ignition, also to clean the target. Also, depending of the diameter of you target, it could be necessary to open the matching box to change the impedance of the coil and capacitors.

First, you should try all that manually before doing an automation.

Cheers

Mikhail,

Without knowing more about the system design, this is a difficult question to answer. My guess however is that it has less to do with your RF generator, and more to do with your gun/target. Are you using a commercially available sputtering gun, or one that you designed and built yourself? How old is your system, and more specifically, your magnets? Conductive targets do not significantly alter the magnetic field as it passes through it, dielectric targets, such as TiO2, will effectively alter/dampen your magnetic field before getting to the ionizing gas (which increases the power required to ionize the gas). My 1st suggestion would be to attempt to use a much thinner target. If you are using 0.250 inch target, attempt to use a 0.125 inch target instead. Better yet, if you have an old di-electric target of any kind, have your machine shop mill or grind that one flat, install that one, and see if you can strike a plasma. Good luck!

Sincerely,

George

We have a system with multiple sputter guns that can be active at the same time. I only managed to ignite RF plasma for a non-conductive target if I had already ignited a plasma in a different gun with a conductive target. So I would start both guns at the same time with the shutters closed, then when both have a stable plasma I switched off the power of the conductive target, then take the pre-sputter time and then open the shutter to start deposition.

Hi Mikhail,

To add to the advise given above I would like you to take notice of the Paschen curve, see picture. In plasma theory this described the minimum voltage at which a plasma ignites. The important parameter here is the product of pressure and electrode distance (p*d) and is shown for different gasses (= different ionization energies). I included such a Paschen curve.  In many sputter systems the electrode distance is in the order of 5 cm. Having said that, even if you have an optimum set for the pressure x distance, oxidic targets tend to ignite worse than metal targets. The information from Pieter Westerik shows this dificulty. Still ignition on oxidic targets should be possible at the right settings, shutters distance to target further than the Debye length and a pressure burst, spark plug or heating filament to introduce electrons for ignition.

Kind regards,

Peter

Yes I think in my case the second sputter gun has a similar function as a spark plug or heating filament: to introduce extra electrons for the ignition.

So far, I have tried:

1) lowering the shutter position relative to the target

2) Pressure burst approach(I also tried to introduce H2 in addition to Ar]- tried up to 10-2 Torr

3) Increasing the power on the target- maximum power I tried was 200 W with limitation of the voltage of 700 V.

4) igniting plasma around another target first(there are 3 target in the system). Also, I tilted the target towards the non-conductive one.

However, no matter what I tried, the plasma on the target has not been ignited and the voltage has always been 0.

I also have an approach with thinning the target, but it seems to be to radical since it will decrease its resource[it is new!]

May you suggest the initial values of the power/voltage on the target and the mode I should try? The available regulation modes are: Power, Delta-P, DC-BIAS and RF-Peak.

Hi Mikhail,

Reading your story it think it is not a normal ignition/process problem but equipment related. The fact the read-out voltage is 0 means that either your RF-supply is not working properly, you have a broken RF-cable, you have bad RF-connectors or you have an RF-leak in the circuit.

1. Checking the RF power supply you can do by directly connecting the generator with a RF-cable to a dummy load and see if you can dissipate power in the load. You should get a power/voltage reading on the generator. Of course you can use the different modes of the generator but this really should not matter much to check if you get dissipation into the dummy load.

Change the RF-cables to check for a broken cable. Also check for bad connectors, oxidised or corroded). Check feedtroughs with a resistance meter to check for shorts to ground.

2. There can be many leaks in the RF-circuitry. The most common are RF vacuum feedthrough leaks were you get contact to ground of the reactor or leaks through the cooling circuit (especially when you have highly conductive or contaminated cooling circuit). It is quite possible that you loose most of the RF power in such a water circuit, depending on the conductivity of the water.

3. Check your matching box connections. Very often you have moveable parts (adjustable capacitors) in here and there might be loose connections. It is also possible the motors which drive the capacitors do not work properly and you cannot get the capacitors in a "plasma start" position. A common problem is that the motor has driven the capacitors to the end position and they are stuck. Check for corrosion (very often the matching box is water cooled) and check if the blocking capacitors are not leaky.

4. Finally check the connection of the RF-cable to the target. A common problem is that the target is not isolated to ground. This can happen if the ceramic or o-ring seals are deposited by metal. You can simple test this by a resistance measurement between target and ground and should read high Mohms. If not, change or clean the ceramic or seals and check if the target does not make contact with ground. I assume your target is bonded to a backing plate. Normally this is a metalbond and there should not be gap between target and backing plate (Cu, Mo). Sometimes conductive polymer bond is being used and this can also result in bad contact. I would advise thermal bonding with (vacuum) solder bond.

Hi, Peter!

I see no plasma when I use the generator on the conductive Ti target. The power was 200 W, the voltage set was 400 V[the parameters used for DC-generator]. The voltage shown was 0 and the power 200 W[after setting the capacitors].

However, the RF-source works with the substrate holder[the etching procedure]. I can see the plasma inside the chamber and the generator reports the DC-BIAS of 130 V with the 50 W and 450 V set.

So, the problem must not be with the cable, but rather the targets. Before each experiment I check the resistance between the targets and the ground. So, it also must not be an issue.

Maybe I need to ignite the plasma using the conductive target and the generator first.

Hi Mikhail,

If the source works for the substrate holder the issue indeed can have to do with the target. Is a different cable used going to the target? If yes, check this one first. Do you use a switch between substrate holder and target? Then please check if you make good contact. You can of course try to ignite with the conductive target but I guess with what you already have done (high pressuree tc) it should have succeeded. Alternatively you could put a different target assembly on the position of the Ti target using the same cables etc. If you get a plasma the system in itself is ok, then the problem must be in the target assembly itself. You mention you see a voltage with substrate holder plasma but not on the target. I can only imagine that you must have a large voltage drop due to some short.

kind regards,

Peter

Mikhail,

Ti targets are not easy to sputter.  700 volts may not be enough to ignite the plasma if the ground that the electrons see is larger than your substrate (something that is nearly impossible to determine, but almost certainly true). Do you have a Au or Cr target, or any other conductive target that is much easier to sputter?  I can sputter Au at 20 Watts, and Cr at 30 watts no problem.  This would at least give you confirmation that your gun is working properly.

Sincerely,

George Peterson

Hi George,

I have metallic Ti target installed in the system now. I can sputter it easily using DC-generator, but it also does not work with the RF-one.

Hi, Peter,

The same cable is used for connecting the generator to the substrate holder. There are the connectors of the same type used for both: target and holder.

However, I did not mean the ignition using another target with Me installed. What I meant was that I could not use the metallic target with the generator. It also showed 0 V. The system has 3 targets installed and I can change cables from the generators easily. So, the problem must be concerned with the targets' assembly. May you suggest what else I can try?

Also, it is not possible to match the capacitors in the matching box so as the reflected power was 0 as in the case with the substrate holder. So, with targets it is always ~20% of the incident power. The mode used in the matching box is manual.

Mikhail,

As I was thinking about your problem more, it occurred to me (again without seeing your specific system) that with your RF source you should see an Ar plasma regardless of your target, as long as it is conductive, if you are delivering enough voltage to ionize the Ar.  It is just a question of if you are sputtering anything off of your target. But you are not seeing that, a few posts earlier, you said that you are seeing a bias of 0 volts with a power of 200 watts. If your system is anything like mine, that suggests you have a short in your gun. What are the chances that you have, or have had in the past, a cooling water leak from a melted or cracked cooling line inside of the gun? 

Sincerely,

George

Hi, George,

I can see plasma near the conductive target with DC source in the same experiment, without changing anything. So, I just connect the cable from the DC generator and can easily ignite the plasma, but when I connect the cable from the RF source, DC Bias always 0. Should I try higher powers?

I managed to ignite plasma with ALL the targets just by opening the shutters. Still, I cannot explain why exactly it worked and I cannot understand how I can pre-sputter.

Hi Mikhail,

First run should be to ignite a plasma with a metallic target in pur argon at low power 100W for example in RF. The DC bias should not be 0V. 

You should test your RF cable from the matching box to cathode connection and to the target surface. Also to check the connection from the target to the RF generator.

If the connection is OK, it could be nice to check the soft parameters of the RF generator , to rotate manually the capacitors inside the matching box.

After you should be in the manual mode to find the good position of the capacitors manually (reducing the reflected power as low as possible) for ignition around 1 ou 2 Pa. You may use also additive capacitors inside the matching box if necessary.

Sincerely

Olivier

here is the possible explanation to the phenomenon of ignition of the plasma in my system

I also managed to ignite the plasma with the shutter close under the pressure of 10^-2 Torr. However, the deposition rate was 0.03 nm/min.

Hi Mikhail,

At high pressures ignition usually becomes easier but deprate drops since you have more gas-gas collisions and you get more dispersion of the neutrals. Do you still have high reflected power at higher pressures (20%)?

Peter, no, as soon as plasma is ignited, I can match the capacitance to completely eliminate REFP.

That's good. If you have initial mismatch and some reflected power, that is quite common. I understood you're doing this manually. But even in automatic systems you very often have a mismatch during the first few seconds after ignition. You mentioned that you use the same cable for both the sputter etch and deposition of the tool. These probably have different surface area's in the system and therefore different initial matching. In the past, when I had problems with automatic matching, I sometimes changed cable lengths to get into an optimum range for the (automatic) matching box. This was just done by trial and error. So, you could consider using two different cable length's, one for the sputter target, one for the sputter etch.