I have sputtered NTO from an alloy target consisting for 99 at. % of titanium and 1 % of niobium. It was reactive DC magnetron sputtering, with a mixed atmosphere of oxygen and argon. The substrate, a silicon wafer with 200 nm of thermal oxide on it, was kept at approximately 300 degrees C during sputtering. The resulting layer of NTO was found to be approximately 100 nm thick (by ellipsometry). After deposition, the substrate was annealed at different temperatures ranging from 400 to 800 degrees C in air. Then, four point probe measurements were conducted to measure the sheet resistance, which was in all cases out of range of the instrument. The instrument does give realistic values for other materials. Then I analysed the NTO by XPS, and it was found that the atomic ratio Ti:Nb was indeed approximately 99:1, and the titanium and Nb were fully oxidized (Ti 4+ and Nb 5+). So what is wrong with it? Even undoped titania should be a bit conductive right?
Pieter,
Your problem might not be the material since you do have two different valence states (Ti4+/V5+) and it is very likely you will have some conduction through electron or hole hopping between these states. It might just be the contact method you use for the measurement. With what are you probing the layers? Do you use Indium contacts or just needles?
I am just pressing four needles in a row on the surface (each needle has its own spring, so all of them are touching the surface). Because it is a four point measurement the contact resistances should not be an issue right?
Pieter
Unless they really are high resistive contacts (schottky?). Maybe an idea to do one test with Indium patches and contact those with the needles
It has low melting point (156C) and by putting a small piece on the samples and heating it on a hotplate it forms nice round balls which are easy to contact. We used to do this on InP, GaAs, InGaAs for CV and Hall measurements. Because of its low melting point it readily diffuses into oxides and III-V's. Alternatively you can try to make a good contact by sputtering (through a shadow mask) Ti, TiAu or Cr onto the sample. I usually use selfmade masks by cutting into polyimide foil or PI tape.
Hope this is of any help.
Yes, maybe it is conductive after all. The layer did not charge in the SEM when we tried EDX on the layer.
I will try to make better contacts by sputtering. I will try first with aluminum because that is what we usually use for electrical contacts to our devices. If that does not work I guess I will try gold with a thin chromium adhesion layer. I will keep you posted with the results.
Pieter,
If it does not charge in the SEM it is conductive and making good contact will probably solve the issue. I think Al is also a good metal to deposit. Depending on the workfunction of the metal you might or might not get a Schottky contact on the NTO. For the 4pw measurements it will probably not be an issue. I would appreciate if you let me know if the problem is solved.
Oh, oh,
it's only today that I saw this question...
Are you aware the TiO2 tends to form crystal needles (depending on thermal conditions - look for Rutile and Anatase @ wikipedia)? These tend to conduct in the z-direction but not so much in x and y. And they are quite fragile when it comes to the contacting with probes. Some supplier of ours tried conductive epoxide glue with some success.
Sputtering metal contacts could easily cover these needles with metal down to the base while still not establishing a conductive path in x and y.
What do you intend to create? Would a metallization layer underneath the oxide growth be feasible?
Regards
I was hoping that the NTO would make ohmic contact to highly doped silicon. This would however be hard to test if the NTO itself is not conductive, so I tried to measure the conductivity of the NTO first. I was however not aware of the anisotropic nature of the conductivity of TiO2. If I still measure no in plane conductivity with the metal contacts I will try to test the conductivity in z direction by sputtering a metal (platinum?) layer underneath and also making metal contacts on the top.
Though maybe covered by some patent: a solid layer of titanium on the wafer side should work. (My guess: annealing in an inert atmosphere will be required.)
As the Ti layer will form an oxide layer on top, an thin (10 nm) underneath layer of Pt could provide a layer for contact attachment. Ti/TiO2 is quite 'tricky'.
The solid titanium should work for what? The ohmic contact or the measurement of the conductivity? Anyway in the end I cannot have a (thick) layer of metal underneath the NTO because it should be quite transparent.
Where exactly do you intend the platinum to be? I did not understand it from your description.
I could not measure any finite resistance between two aluminum contact pads on top of the layer. My next plan is to measure through a stack: Au/Ti/NTO/Ti/Au, or maybe with platinum instead of gold.
Hi Pieter,
That's very appointing Al as a contact layers does not work. It could be related to work function so choosing other adhering materials would also be my idea (Ti, Cr, Nb, Mo, Ni) certainly in combination with a ductile contact layer of Au on which the test pins will make good contact.
Sorry for the late reply...
About Pt/Ti base layer:
If you insist on your substrate (Si/SiO2), could be that a Ti layer does not adhere well to the oxide. If so, Pt could be used to act as some 'glue' - bonding well to the oxide as well as to the Ti. The Pt layer could go as low as 3 nm (not well-defined but working).
The Titanium was intended to
Regarding the Al pads: As already stated I doubt that you can get some resistance parallel to the wafer surface. Due to the needle structure of the TiO2 I expect conductivity - if at all - only in the z-direction. (Read 'if at all' as 'if everything else goes well'.)
Regards
Thank you for the clarification. As far as I know, titanium adheres better to oxides than platinum, since it has a higher reactivity. So titanium should adhere well to most materials. However the conductivity might be an issue if it is oxidized. That is why I propose for my new test to encapsulate the whole stack between noble metal layers. I will use platinum and not gold, since the technician of our reactive sputtering system prefers not to have gold in his machine.
Also, the new test will measure the conductivity in the z-direction, so if this test gives no conductivity I think I can safely conclude that my material is not conductive in any direction.
- Badges
- Science topic
- Similar topics
- Chemistry
- Nanotechnology