I am investigating the multiferroic properties of doped Bi4Ti3O12 thin films on platinum coated silicon substrates. As part of my investigation, I am looking at minimising the film thickness. One of my initial experiments on thickness control involved using cross sectional SEM to measure the film thickness of samples deposited for different time periods, but otherwise processed using an identical methodology. Samples deposited for 1 hour, 2 hour, 3 hours and 3hours and 18mins(control sample) all show ideal grain flat grain boundaries between the BTO and platinum, and also the platinum, titanium oxide and silicon. However the thinnest sample, deposited for only 30 mins, shows a roughened substrate, as seen in figure 1.
At first, I believed this may be due to damage when the sample was cleaved in half, but another phenomena occurred later.
To investigate the electrical properties dependence on thickness, I fabricated thin 4 films using the same process again, with deposition time controlling thickness. The films were 50nm (24mins), 100nm(49mins), 200nm(98mins) and 380(186mins) thick . Gold contacts were sputtered onto the BTO surface, so electrical contact could be made between to the top and bottom of the film by placing probes on the Au and exposed Platinum substrate.
As seen in figure 2, an almost linear relationship can be seen between 100nm and 380nm, however a huge drop can be seen in the 50nm Sample.
A drop in capacitance could be explained by an increase dielectric contribution between surface states, but roughening of the substrate as seen earlier could also affect the crystallinity of the film.
I plan to performing cross sectional measurements to confirm the roughening, but am still unable to explain this why this occurs. Could anyone explain why the Platinum films roughens specifically when a thinner film is deposited on it?
Best wishes
David Coathup
Two important details are necesary, before analysing the situation.
1) what are processing temperatures for your bismuth titanate thin film.
2) What technique are you using to deposit the ferroelectric thin film.
3) Are you using platinised silicon substrates.
4) The cross-sectional study done by TEM, will be more useful, and reveal much finer details, rather than SEM.
5) You mention capacitance, instead of that, can you calculate what is you measured dielectric constant for the deposited Bismuth titanate thin film, and at what frequency are you measuring the capacitance.
Kindly let me know the abvoe details.
K. Sreenivas
Thank you for taking your time to help me in this matter.
To list of the answers to your questions.
1) The samples were annealed at 650OC in ambient atmosphere
2) I am using RF sputtering with a pure Ar atmosphere and no substrate heating
3) The substrate is silicon with 5 µm of silicon dioxide, 400nm of titanium dioxide, and 150nm of platinum on the surface.
4) I will enquire with are technicians about the details of using TEM for this purpose
5) I have potted the dielectric constant vs thickness in the attached figure.
The dielectric constants for 50nm, 100nm, 200nm and 380nm are 18, 161, 173 126 respectively. The values between 173 and 126 are comparable to that of compared to literature. The peak at 200nm I believe can be attributed to a balance between two factors. An increase in the dielectric constant of the bulk of the film at smaller thicknesses, and the ratio of film bulk and surface states.
Many thanks for your assistance
David
Dear David,
You must check a lot of things systematically.
1) RF sputtering an oxide target (Bi4Ti3O12) in Ar, would normally give you an oxygen deficient film. Sometimes Bi deficiency can also occur. You must check the XRD patters of all your films i.e., all the various thickness films, and ensure that your XRD of the annealed films is close to that of bulk Bismuth titanate, you will see a polycrystalline structure, and should see a good pattern. This ensure the formation of a single phase film.
Try to compare film sputtered in Argon, and AR+O2 sputtering gas as well.
2) Now you mentioned annealing the film at 650 deg C. Just take you bare si/TiO2/Pt substrate without the Bi4Ti3O12 deposition, put it in a furnace, and simulate the same kind of thermal treatment and atmosphere, and see that the electrode substrate is stable. Stability means;
i) the Pt electrode is conducting
ii) The adhesion of the electrode is retained.
iii) the surface has not become rough.
iv) Try to compare the annealed and the un-annealed platinized silicon substrates
Look at the surface micro-structure with an AFM. I found the atomic force microscope revealed the changes in the electrode surface very well in comparison to normal SEM.
this will let you know in case the platinised substrate is getting disturbed or roughened while you are processing/growing the Bi4Ti3O12 film.
v) You are seeing a Bi4Ti3O12 film thickness dependent dielectric constant. At higher film thickness the measured dielectric constant is close to the bulk values.
At lower film thickness the values appear to be quite low. this can be due to a variety of reasons (crystallinity, grain size, presence of any interdiffusion with the electrode etc.).
Dear Professor Sreenivas,
thank you very much for you continued support with my work, I am incredibly grateful for your insight. I am aware of the difficult of sputtering from a stoichiometric target, however due to are sputtering apparatus using a cryopump, I am unable to sputter using a O2/Ar gas mixture for health and safety concerns, however I been able to overcome this issue with the annealing in ambient atmosphere.
The films I have produced show good ferroelectric properties, albeit slightly less than that recorded in the journals. This drop could be due to the novel doping, but is also likely to to the lack of optimization in film thickness and post deposition environment, the latter of which will greatly effect the oxygen vacancies in the film.
I have access to AFM, and will attempt the experiment you mentioned. A theory I have discussed with a college is that the absorption of atmospheric impurities into the Pt or the layers beneath may cause the roughening. However, thick BTO layers prevent his, which is why the roughening only occurs in the samples with the thinnest BTO layers. Does this sound pausible to you?
Many Thanks
David Coathup
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science