How did you conclude that the films are TiO2?

Getting oxidized Ti during sputtering is not surprising. The oxidation occurs most likely during the deposition of Ti, either by residual oxygen in the vacuum or insufficiently pure Ar, leaks in the gas manifold etc. Ti has a very high affinity for oxygen and it will grab the smallest amounts of residual oxygen in the vacuum or gas mixture. 

Additional possible sources of oxygen, although I believe that they are less likely:

Hi Artur, Lior!

The time for the deposition of 70 nm of Ti is around 5 min. However, the deposition speed of TiO2 is much slower, around 1 nm/min. After depositing the Ti layer, its color is metallic and it is almost opaque, but after depositing TiO2 layer, it is black and transparent.

In addition, in order to convert 70 nm of Ti into pure TiO2, one will need 2000+ nm of TiO2 deposited on top of the Ti layer.

I have prepared other species of TiOx before(Ti2O3, TiO) and I can distinguish one oxide from another. In addition, I confirm it by UV-VIS spectrophotometry.

I am also thinking it could be oxygen in Ar[used for the deposition process], but during annealing, the atmosphere is vacuum, 10-6 Torr.

Hi Mikhail,

How do you measure the deposition rate or thickness?

I still find it hard to understand how can you distinguish between Ti and various TiOx phases. I read your comment twice and it sounds as though you're identifying the phase/stoichiometry with your eyes? I'm probably missing something here, since this type of characterization problem isn't usually straightforward to address.

As to the main question: the most likely candidates for your oxygen is the gas line, + even if the Ar is pure, 10E-6 Torr isn't sufficient in my experience to obtain metallic Ti. 

Titanium will suck the residual oxygen rapidly (it's actually used in TSP pumps for this exact reason). When I started depositing Ta, which in my experience is less reactive to oxygen compared to Ti, I got 20%(at) O when depositing at 1E-7. I solved it by reducing the base pressure to the low -8s and optimizing other process conditions. Again - for Ti the problem would be much worse and you're at the -6s. Same goes for the vacuum anneal. To make things worse, your deposition rate, assuming that you measured it correctly (~2 A/s) is relatively low and this would give the fresh Ti layer on the surface relatively large time to get oxidized before it's covered with more Ti. In some cases, increasing the deposition rate can help reducing the oxidation.

Also, in sputtering you cannot assume that your old process conditions are kept constant over months; the target can change its geometry, composition (get oxidized) etc. This means that your previous possible success with growing metallic Ti might not be the case every time your grow, the deposition rate may drift considerably as well.

Hi Lior,

I confirm my visual inspection with XRD, UV-VIS spectroscopy and Raman, but not all the time. I know that TiO is bronze, while Ti2O3 is blue or purple. TiO2 is transparent and oxygen-deficient TIO2 is semi-transparent and black. Magneli phases are black or dark-blue.

May adding 1-3 sccm of H2 during the deposition/ annealing of solve the issue?

Hi Mikhail,

Normally the O2 partial pressure in PVD chambers is 2 orders of magnitude lower than base pressure. So at 10-8 T O2 it should not be a problem to sputter pure Ti layers.

I know from experience that if you have Ti and TiO2 targets in one vacuum chamber the Ti target can get contaminated when your sputtering TiO2. Oxygen from that target will be gettered by the Ti target and oxidize the surface. If you want to sputter clean pure Ti you need to clean up the Ti target after every run. A dummy run prior to the Ti deposition can help you solve this issue. Of course you should not have air leaks in your chamber and the Ar gas should be of 5N quality. If you doubt the vacuum integrity of the chamber hook-on an RGA an do a gas analysis.

With a Ti-target clean-up you should be able to deposit 70nm Ti followed by 120 nm TiO2. As mentioned by the comment of others Ti is extremely sensitive for oxygen and I would no be surprised if during the anneal Ti is oxidised because  O2 is present in de anneal chamber or because of some intermixing. A 120 nm TiO2 layer is optically not completely closed and usually you have some columnar TiO2 growth so that there will be voids in that layer. The TiO2 layer will therefore not protect the Ti from oxidizing.

In the past we used Ti-layers on microscope glass plates to check rapid thermal anneal equipment for leaks. If you have the slightest oxygen traces in such RTA the Ti will already start to colourise after 20s rapid anneal at 700C.

For how long do you do the anneal in (I pressume) an oven? Consider doing this in a RTA with Ar-flow. This will help you to prevent oxidation.

Peter

Mikhail,

To summarize my previous and following ramblings on thin film characterization in one sentence: My experience has taught me repeatedly to be very skeptical and suspicious of any unconfirmed parameters of films (thickness, composition, structure et al.)

(as my mentor once said during a debate of this nature - "I'll believe you when you show me the TEM cross section of it" - and he was spot on)

These parts usually don't make it into the experimental section of the paper, but I could share at length the countless times I saw my own and my colleagues' assumptions shatter into reality after spending time, effort and resources that could have been conserved if said assumptions were challenged on time.

While in general your color coding is agreeable for bulk material, I would be very careful in putting any weight on such observations for thin films. The color and transparency of thin films is significantly affected by their thickness,* which is another parameter that I don't see you having reliable measurements of (particularly for multilayers).

Again, changes in your target (thickness, geometry, stoichiometry/contamination/oxidation) can dramatically change the deposition rate between growths.

Just for example, you cannot use XRD to determine that all the Ti you deposit is oxidized. The Ti could be amorphous, it could be very thin, or lay between partially oxidized particles/precipitates and other possibilities.

An idea for a simple estimation of the presence of some Ti would be to scribe two corners of your film and touch them with an Ohm-meter. Having an insulating result would strongly suggest that indeed no Ti is metallic.

The reason I'm dwelling on your characterization issues is simple: the problem may not be what you think it is, so understanding it would help solving it. This should also be addressed on a broader perspective: if you only need this film as some electrode/coating and it needs to fulfill a simple task, then you might not be interested in this failure analysis. Conversely, if your application depends/assumes non-oxidized Ti or is otherwise dependent on / sensitive to the precise film parameters - I suggest taking a more thorough look into the structure and chemistry.

As for your question, adding H2 could help mitigate residual O2 - depending on its source and severity, but it could also affect the sputtering rate etc. 

* see attached example from DOI: 10.1039/C4DT00711E for color versus thickness of several materials

Dear Peter,

I deposit 70 nm of Ti layer first and later I do not use this target, only TiO2. Also, I cleaned the surface of the Ti target before placing it to the chamber and I pre- sputtered for 30 min to eliminate any possibility of oxidation the target before/during the deposition. I also can confirm that Ti layers deposited first are real metals. I attach a picture of the pedestal with the samples.

After deposition of the TiO2 layer, I annealed samples for 12 h at 700 C in vacuum. I believe that the time used for annealing is the most crucial parameter affecting oxidation of the Ti layer since after it has been just deposited, it is metal. However the deposition time is minutes, while depositing 120 nm of TiO2 takes hours. Also the annealing step requires many hours more. So, if there is a source of oxygen in the chamber, Ti layer has a lot of opportunities to catch it.

We also have RTA, but it is another story and it was not possible to keep the oxygen content constant during annealing in our RTA due to the leakage.

Dear Lior,

We have obtained quite a few TiO2 samples colored into different colors(from yellow to red), but they all were semi-transparent. The TiO,Ti2O3 and magneli phases samples are opaque. We measure thickness in each experiment. For that, we protect some regions on the samples with correction fluid and then, after the films has been deposited, we remove the fluid with alcohol and measure the thickness of the film deposited with an optical profilometer. Attached is a slide on which there is a sample with two layers one on top of another( Ti layer deposited in the first step and TiO2 layer deposited in the second step). There is also a photo after an annealing operation at 500 C for 2 h.

Next experiments I am going to conduct in this branch is to introduce H2 during the process in attempt to compensate the harmful influence of a mystic source of oxygen in the chamber. I also will conduct experiments of oxidation of Ti layers in the chamber without oxygen introduced on purpose.

Hi Mikhail,

Your statiuing that the Ti film comes out the sputter system asd metallic, Considering the very long anneal step at 700C I think it might be a good idea to just put a Ti sample without the TiO2 layer in the anneal chamber and do an anneal to see if the Ti oxidises. If this is the case I would not anneal in vacuum but with using 5N Nitrogen flow to avoid oxygen coming in the system. 

This is a nice and most appropriate suggestion to check what is happening during annealing.

May be the oxygen is diffusing from your upper TiO2 layer towards the below lying Ti layer and hence making the whole TiO2.

Mikhail,

I suggest trying the conductivity measurement I suggested before and after the anneal (scribe again before each measurement). If you lose conductivity during the anneal then the growth conditions are fine. I agree that the anneal is a likely cause for oxidation. Is this high thermal budget mandatory for your end goal?

If it is - I would try the hydrogen during anneal as well (forming gas).

Also, and this requires 5-15 min in Google Scholar - I'm pretty sure that (a) in these anneal conditions metallic Ti would reduce SiO2 to form TiOx or at least form silicates (b) there must be many papers on this.

Let us know what you come up with

Good luck,

Lior

P.S.

See for example the excellent work by J. Carbal @IBM Yorktown showing some phase formation with Ti/SiO2 at ~400C

Journal of The Electrochemical Society, 151, F283-F287 (2004)

http://jes.ecsdl.org/content/151/12/F283.short

while annealing at 700 C the oxygen may diffuse from TiO2 layer to into Ti layer making all film TiO2..

I too suggest that the O2 comes from the TiO2 Layer. To avoid that, you can try to use much lower annealing temeratures and shorter times. I would try 300 - 350 degrees and 30 min - 1 h. On the other hand you can sputter a diffusion barrier between Ti and TiO2. For this you can use Pd for example. 

Hello Mikhail,I think: 

Comments are true. For further diffusion phenomenon, you do evaluation of percent oxygen Before annealing process in TiO2 layer by methods analysis and use a non-titanium layer ( Ni, Al or each element like titanium in terms of crystallographic latice and diffusion ) for study of oxygen diffusion into the middle layer under the annealing process.

I think:

Theoretically, Ti should not be able to act as reducing agent for TiO2. However, It might be possible for Ti to reduce SiO2 and get oxygen from there to make TiO2.

It does not make sense that when the annealing is being done of Ti and TiO2 then Ti becomes TiO2 without reducing SiO2..

Asif Ur Rehman , why is not it possible theoretically?