Bi and Bi1-xSbx Project
Created by ESW, April 2018
Background
Started by Prof. Akinwande as a collaboration intended for 2-D electronics (high mobility of Bi + semiconducting due to quantum confinement).
Small band gap + surface states makes electronic device applications difficult
Transitioned to focus on materials science & spintronics (high SOC, surface states --> SHE and SOT)
Transfer focus was accidental - observed films delaminating during TEM prep (w/ Prof. Larry Lee's group).
Important points
In bulk crystals, BiSb between 7-22% Sb is a topological insulator with two surface state bands and a total of 5 Fermi level surface state crossings. This range may be expanded through quantum confinement, but this has not been confirmed experimentally.
Bi is not a topological insulator unless it is 1-2 bilayers thick (in the (001) A7 structure). We cannot grow Bi this thin on Si(111), because Bi forms a puckered-layer A17, (012)-oriented structure in the first 4 nm of growth on Si(111) before transitioning to the A7 structure (see Nagao papers). Quantum confined Bi (<100 nm) has surface states that have spin orbit splitting due to the Rashba effect. TIs have "spin momentum locking" in the surface states, which makes the spin polarization potentially more robust.
The A17 structure may have a larger band gap, but experimental studies are very lacking. BiSb A17 has been theoretically predicted but not confirmed experimentally. I think we are probably growing it, but we need to protect the nanoscale films better from oxidation to get more reliable transport measurements of those films. A17(012) and A7 (012) have the same d-spacing, so cannot be distinguished in XRD - need to be distinguished through in situ ARPES (we don't have this) or through distinctly different transport.
Practically, TIs are distinguished from non-TIs with conductive, spin-split surface states through ARPES - TIs have an odd number of Fermi level surface state crossings, and non-TIs have an even number (i.e. Bi has 4 crossings). To the best of my knowledge, you cannot distinguish between a TI and a non-TI with conductive surface states through magnetotransport measurements.
Weak antlilocalization (WAL) is not equivalent to surface states, even if the alpha prefactor is 0.5. WAL can also arise from SOC in bulk states. You need to consider the thickness dependence to determine if features are coming from the surface or the bulk.
Sb alloying with more than 4-9% Sb suppresses the A17 to A7 transition, resulting in (012) oriented bulk-like films that probably have the A7 structure. Growing the BiSb on a Bi(001) buffer layer seems to prevent this, but having parallel Bi and BiSb layers is likely to significantly complicate magnetotransport and conductivity measurements unless planned very well.
Growth Procedures
Growth Procedures
Calibration
Points to remember
LC bake at 450 to avoid desorbing hydrogen passivation in the load chamber
Always use a Si backing wafer (we have confirmed experimentally that sapphire wafers will shatter at the deox temperature).
Because the growth temperature is so cold, remove wafer from GC immediately after growth to avoid crap settling on surface
The melting temp of Bi is ~270 C. Do not anneal Bi above this temperature following growth, or it will desorb.
As capping Bi is a bad idea because 1. As can alloy with Bi, 2. As is a semi-metal, which will seriously complicate transport measurements, and 3. You cannot desorb the As thermally without damaging the Bi.
Heating the Bi at ~100 C after transfer to an XPS chamber (UTD) seems to clean up the oxide, but this isn't exceptionally well-documented.
I think Al capping and letting the Al form an oxide layer is the best route toward protecting the thin films going forward with our current set of source cells.
Ongoing collaborations
ARPES (Prof. Kawasaki)
Current status: Alec and Stephen grew 50 nm BiSb wafer targeted at 7% Sb (B180515A) on conductive Si(111) substrate
Need to cleave it and mail to Prof Kawasaki
No precrack /DCB prep needed
In situ transfer for ARPES tentatively scheduled for June
Goal: measure band gap - is it larger due to quantum confinement? Is the surface state dispersion different in thin films compared to bulk? Compare with H. Benia paper.
Transfer (Prof. Liechti & Prof. Lee)
Give DCB samples of 80 nm Bi or 50 nm BiSb(001) (B180515A) to Tianhao. He will repeat DCB experiments to figure out if EP30 epoxy is still working.
Mail samples of transferred Bi on epoxy (by DCB method) - already in transfer desiccator from previous experiments with Seung - to Prof Larry Lee for TEM looking at how delamination happens at interface
Grow different thicknesses of Bi and measure how adhesion depends on the film thickness
Theory and Antimonene (Prof. Palacios)
Antimonene
Grew 10 nm and 4 nm Sb at substrate temp of 220 C TC, very roughly approximated 270 C pyro
10 nm film is mixture of (001) and (012) orientations
Do XRD on 4 nm film to check if orientation mixture is thickness-dependent and XRR to check thickness
Do AFM to check island size / shape
Ideally would get ~ 2 nm Sb (001) for electrical measurements by his collaborators
Would be cool to explore Sb growth generally, as this hasn't been looked at much on Si(111) yet, unclear if puckered-layer antimonene can be grown
Might need to protect from air exposure
Prof Palacios (and physicists generally) prefer to use the rhombohedral unit cell for Bi, so they use (111) for (001) and (110) for (012). Both are equally valid. The trigonal coordinates make it easier to interpret XRD
Bi Theory
Preliminary band structure simulations suggests Bi surface states are not topological but are still ok after oxidation
ESW - send conductivity, hall, MR data on different thicknesses of Bi(001) to Prof. Palacios's post doc
They will try to simulate temperature dependence and oxygen coverage and compare with experimental data
May request conductivity measurements of a sample protected from oxidation. Options to do this include spin-coating with PMMA immediately after growth (easier, but may have some O2 exposure), or capping with Al in situ and allowing Al to oxidize (requires calibration of Al deposition rate, but can cap in situ). Be careful about unoxidized Al remaining in parallel with the Bi.
MOKE (Prof. Incorvia)
Gave thickness-dependent Bi and BiSb(001) films to Suyogya
They will make contacts to them with lithography in a transistor-like structure, and plan to start MOKE measurements in June
May request additional samples after June
SOT (Prof. Salahuddin and Niklas Roschewsky)
STT measurements of BiSb(001) film indicate that previously reported unusually large SOT is actually anomalous Nernst effect
Niklas is writing the paper
May request sample data (XRD, AFM etc), ESW can probably handle it
Magnetotransport (Deepyanti Taneja, Tanuj Trivedi and Prof. Mason)
TBD
Materials Characterization
XRD
Short-range omega 2-theta scan:
Range: 8 - 15 degrees in omega
Step size 0.003
Speed: 0.3
Purpose: shifting of the Bi(003) and Bi(012) peak. It needs to be quite wide (down to 8) to make sure you catch the (003) peak of even very thin films, which can be quite broad.
Takes 1 - 1.5 hours
Wide-range omega 2theta scan:
Range: 5 - 60 degrees in omega
Step size 0.005
Speed: 0.5
Takes 2 - 2.5 hours
Notes
These scans may seem slow. It's necessary to use these settings to resolve very thin films, especially in the nanoscale Bi. Additionally, since my samples degrade gradually over time, sometimes unexpectedly, I prefer to get a high enough quality XRD image to use in a conference talk or paper in the first pass
In BiSb the composition varies radially. Cleave a small (~ 1 cm) section from the center and use that for XRD. Not necessary for pure Bi or Sb (the thickness variation appears within 1 nm, which doesn't affect XRD measurements too much). The crystal structure of BiSb is very sensitive to small changes in the composition, particularly around 8-9% Sb
More notes on transferred sample XRD TBD
AFM
To look at surface roughness of a general Bi or BiSb film, assumed to be a continuous film, I usually do one 10 um x 10 um scan to look for Bi droplets and wide-range surface roughness, and one 5 um x 5 um scan to look at morphology (i.e. Bi(001) films have a triangular morphology, Bi(012) films are more square-ish). Please use your best judgement about whether you need to do both type of scans for all samples - just make sure that all scans you're planning to directly compare are the same size.
For likely island-mode growth (i.e. the Sb films), I do a 2 um x 2 um scan on the clean room AFM (digital instruments #2). If the islands are too small, they will not be able to be resolved with the clean room AFM (because the tip is ~ the size of the islands). So far the Sb films have been able to be measured with the clean room AFM, but the Bi(012) films are too small. If you cannot resolve the islands, please contact ETY's group for high resolution AFM. I usually do 1 um x 1 um scans with their AFM - please double check the AFM images in Chapter 6 of my thesis (now on the share drive) to see what images I have already. Chris Brennan did this for me for my PhD, but he is also planning to leave soon, so may need to coordinate with someone else.
Alternatively, we could purchase sharper tips for the clean room AFM and change them out each time. You would probably need more than a 2 hour reservation to do this. Please contact Chris or ETY for details about the current size tip we're using. You need extra certification to change the tip, please contact Sarmita.
The y-axis (height scale on AFM) is flexible, and can be changed after measuring in the digital instruments software. I usually use 5 nm or 10 nm. Please refer to my previous measurements to make sure all images you're planning to compare (i.e. all Sb film measurements) are on the same scale.
For instructions on AFM measurement and tip change in general, see AFM