For this webinar we teamed up with Prof. Martino Poggio, whose work focuses on understanding the physics of materials and phenomena that is strongly dependent on the local conditions, be it by impurities or through the design through local probing of magnetic fields coming from currents and magnetization. Several notable examples include superconductivity and magnetism in 2D material's, spin structure and phenomena in topological insulators, and spin ice. For these phenomena to be understood Martino's lab employs several techniques such as the scanning SQUID magnetometry, magnetic nano-wire NW microscopy. Zurich Instruments' lock-in amplifiers proved to be an invaluable asset through high sensitivity, high resolution and modularity that enables multiple signal detection and control at the same time. In particular, the MFLI with the MF-MD option enabled measurements of the current and the dissipated heat in a sample concurrently using the SQUID magnetometer thanks to the 4 demodulators of the MF-MD option (read more about this option in this blog post). Furthermore, adding the MF-PID option enables tracking of two resonant frequencies of a magnetic nanowire at the same time, thus allowing the measurement of in- and out- of plane components with this method.
In the Q&A session we addressed several questions, which are answered here by Martino.
On the imaging of SC qubits (slide 24): Is the scanning SQUID method frequency-dependent? Whats the bandwidth of currents that are measurable? Those SC qubits are usually operated at microwave frequencies, does that introduce difficulties?
In principle the bandwidth of a SQUID can be extremely high, enough to see the HF currents in the qubits. In our scanning SQUIDs, however, mostly because of the amplifying eletronics, which are not made for HF operation, we work below 100 kHz. So, we can only look at low-frequency currents, which -- as the questioner suggests -- may not be indicative of all currents used in the qubit experiments.
Is it possible to get the references Martino had in his slides?
The references are in the slides. I would be happy to send them to this questioner.
How do you calibrate a SQUID detector? Do you use a known pattern that you can measure with something else?
The SQUID produces a signal that is periodic as a function of the magnetic flux (magnetic field times area) threading through it's loop. The period is the flux quantum (constant of nature). By applying a known field, using e.g. An external magnet, this periodicity can be used to extract the exact area of the SQUID loop. With this information, all subsequent field measurements are calibrated.
How do you handle NWs as scanning tip? How long does such tip last for a typical experiment?
NWs are surprisingly robust and resilient, because they typically are long and thin and have small spring constants. This means that when they 'crash' on the surface, unlike stiff AFM cantilevers, they do not exert a very large force and simply bend back. Therefore, NW are rarely damaged, unless there is some catastrophic loss of the scanning probe control. We tend to use the same times for multiple experiments -- for months even.
How easy/difficult is it to place NWs on scanning head?
There are various methods for placing NWs on a scanning tip and they vary in complexity. Some NWs can simply be picked up with micro-manipulators under an optical microscope and glued to scanning tips. This can take for 10s of minutes to a few hours. NWs can also be grown directly on scanning tips by focused electron beam induced deposition (FEBID). This requires a special SEM and hours of work. Otherwise, NWs that are growing directly out of a substrate can also be used. The substrate needs to be cleaved so that there are NW near a corner of the chip and these can be attached to the scanning probe system and used directly. This can take 10s of minutes to set up.
Is it in principle possible to put the NW tip to the end of the STM tip or MFM setup?
Yes, there is quite a bit of work to functionalize AFM and MFM tips with NWs to increase spatial resolution. This has been done for the last 15 years or so an has allowed, e.g. for 10 nm resolution in MFM. This is also possible for STM with conducting NWs. I am not aware of such work, but it should be equally possible, but it's not clear to me what the advantages for STM would be.
There are industrial applications for sensitive magnetic field imaging in harsh environments at elevated temperatures (say 100 degC - 150 degC) such as eddy current inspection for corrosion in underground steel tubes. What would be the challenges for NV center based sensors at elevated temperatures? Which technique would be most appropriate for these elevated temperatures?
Certainly of the three I presented, SQUID microscopy is not the one to use because of it's low Tc. MFM may work and there is a lot of AFM at elevated temperature, e.g. to inspect wear on engine parts. NV is probably the best as it work well in ambient conditions and does not require vacuum. Unfortunately I don't know what the challenges are at high temperatures, not being an expert on NVs.
Can you please comment on the dynamics for these and other techniques available?
The scanning SQUID is limited to 30 kHz of bandwidth due to its amplification electronics. In principle, the SQUID itself could have GHz bandwidth, but not in our current setup. The scanning SQUID images shown in the talk take about 1 min or so to measure. So, some dynamics can be seen, like the hopping of superconducting vortices in MoSi, but these are slow with time scales in the seconds regime. Evidence of faster dynamics can also be see as in this scanning SQUID-on-tip paper by Zeldov, but averaged over time.
The scanning NW MFM uses transducers with resonant frequencies on the order of 500 kHz or 1 MHz and quality factors on the order of 30k. This means that they have time constants on the order of 30 ms. To measure faster dynamics, the system can be damped by an electronic feedback loop. How much you can damp, depends on the SNR of the measurement and ultimately on the detection efficiency of our interferometric scheme.
We have not tried this, but I suspect using this method dynamics on the 1 ms or 100 us scale could be reached. In the scanning measurements shown, we did not push the time-resolution. 2D images again took on the order of 1 minute, similar to the scanning SQUID experiments.
What's highest resolution could be reached using the nitrogen vacancy magneto spectroscopy? Is it possible to measure the shallow magnetic field from the twisted bilayer graphene using it?
NV centers have recently achieved 100 nT/SqrtHz sensitivity and spatial resolutions from 25 to 40 nm. Nevertheless, as discussed in the talk, this is not the best technique for measuring currents in 2D materials such as twisted bilayer graphene. Scanning SQUID is much better at sensing magnetic field produced by currents, albeit at worse spatial resolution. If the goal is rather to measure magnetism from twisted bilayer graphene, then NV is probably the right choice. In fact, Thiel et al. used scanning NV successfully to measure the 2D magnet CrI3.
Can this technique be implemented in a liquid medium?
MFM and NV should be possible in liquid, but with worse sensitivities that what was discussed here. Scanning SQUID is not possible in liquid, unless we are talking about liquid helium.
What is the ultimate objective your company/research group is trying to achieve with these techniques?
We are ultimately interested in deciphering the mechanisms behind a variety of new and poorly understood condensed-matter phenomena. Macroscopic manifestations of quantum mechanics involving strongly correlated states, e.g. superconductivity and magnetism, are sensitive to the local environment. In many cases, nanometer-scale spatial resolution is required to investigate and identify the conditions for their emergence.
Are you offering PhD students to visit your lab to learn these techniques to work on 2D layered materials in this COVID situation? Obviously should be asked personally. Just I am curious!
If you are interested in our work, contact me directly.