SPM User Meeting at the NC-AFM Conference

With this 7th edition of our SPM User Meeting, we hit two new milestones: 

• For the first time, our in-person meeting were held in the Americas on McGill campus in Montreal, Canada.
• This meeting was held as a satellite event to a major SPM conference, namely the Noncontact-Atomic Force Microscopy conference (NC-AFM), which has traditionally been at the forefront of new advances in AFM instrumentation.

Since this event was held in a hybrid format, the talks and tutorial were recorded and are available on our YouTube Channel. It was great to trigger so many discussions and challenge new ideas, both during this workshop and after the NC-AFM conference. Thanks to all for joining us in person as well as virtually!

Highlights of the Talks and Q&A

Since the questions segment is cut from the YouTube video, we'll report the answers here as well as online questions that could not be answered live.

Resonance Enhancement Techniques in SPM

Backgound: Dr. Romain Stomp discussed Signal-to-Noise Ratio (SNR) and how to improve it via resonance enhancement techniques. Q-control method and multi-frequency examples were also provided.

Q&A session:

• Q1: Q-control should actually increase noise, even with higher Q-value. How come the frequency shift noise is lower with higher effective Q? 
• A1: It’s true that any feedback loop introduces noise and therefore also Q-control together with a phase-locked loop. The way I look at it is that the Q-control changes the open-loop characteristic of the resonator for the PLL to work more efficiently. This would however need further investigation, and a more systematic approach will be performed in a follow-up blog on this topic.

• Q2: Did you try an alternative method to compute amplitude and phase other than lock-in?
• A2: From an FPGA-based digital implementation, we could indeed implement any mathematical algorithm to make similar measurements, but it turns out that the lock-in amplifier method provides the best compromise. We are not ruling out other algorithms, and it’s true that for high-speed amplitude and phase measurements, there can be a better method, but we focused on high reliability and noise rejection cases.

• Q3: Is it possible to do Q-control using the basic HF2LI lock-in amplifier?
• A3: Q-control works with any Zurich Instruments lock-in amplifier, therefore also with the HF2LI, but it requires the HF2LI-MF and HF2LI-PID options to perform the feedback on the X and Y components simultaneously, as shown during the demonstration.

• Q4: I want to evaluate power noise level of deflection signal before and after lock-in in mV/sqrt(Hz). How could I do it correctly? Should I use the plotter, spectrum, or sweeper module? 
• A4: All those modules may use different averaging, filters, settling time and statistics. The best way to measure noise is either with the Spectrum Analyzer tab in Spectral Density mode for FFT(X+iY), or with the Sweeper in Noise Amplitude mode. The latter is able to do it over a wider frequency range. Both can provide you with value in V/sqrt(Hz). The plotter can provide you with good SNR value but for a given bandwidth, not normalized in sqrt(Hz). For noise measurement before the lock-in, you should only rely on the Scope/FFT module.

• Q5: Does this Q-control technique still work when the native Q-factor reduces a lot during measurement? I have a lever with Q-factor around 30k but during measurement it reduces to 10, and I want to know if Q-control can help follow the resonance, if at all, for a different value of Q.
• A5: From 30k to 10 seems like a lot, and I would need to better understand the experimental conditions. But with a Q of 30k, you could easily use a PLL and AGC to track the frequency and measure the dissipation which should therefore keep the loop in a steady state, unless you have some very large non-linear effects. You may further use the Q-control to decrease it to let say 10k to ensure better stability.

Exploring the full potential of multimodal cantilever for MFM, KPFM and true atomic resolution AFM

Backgroud : In his talk, Prof. Hans Hug review the advantage of cantilever based mulimodal and multifrequency operations with specific examples for Magnetic Force Microscopy (MFM) and Kelvin Probe Force Microscopy (KPFM). He also explains the advantages of integrating Zurich Instruments lock-in amplifier and PID for the most advanced SPM techniques.

Q&A session :

• Q1 : For capacitive feedback at 2-omega, the dielectric constant (Hamaker constant) also affect the electrostatic force, how does this plays a role in the topography?
• A1 : Thanks for your expert question ! Indeed, the capacitive topography feedback works very nicely with metallic surface. For our magnetic sample with a metallic cap layer that worked perfectly. For heterogenous samples, the capacitance will indeed change (with various doping profile for instance) but there exist other methods to measure more accurately local capacitance from higher frequencies up to microwave that could provide some correction to the topographic feedback.
   
• Q2 : With rather soft cantilever, did you need to use bimodal techniques to avoid snap it to contact ? What are the criteria for your feedback stability? 
• A2 : For rather soft cantilever, to prevent snap into contact, you need to oscillate at sufficiently large amplitude. That’s indeed one advantage of using bimodal excitation where the first amplitude is sufficiently large to provide stability so that the second mode can have lower amplitude. This was shown with rather stiff cantilever, but it also worked well with our soft cantilever. The second criteria is of course the PLL and the dissipation loss feedback to prevent instability when the amplitude drops.
   
• Q3 : What is your fundamental detection limit in term of kT ? How stiff needs to be your cantilever in order to be not limited by the instrument?
• A3 : For soft cantilever, the thermal noise is the limiting factor while for stiffer cantilever and even more so for tuning fork, it is more limited by detector noise. For our optically detected cantilever, it’s simply the thermal noise of the cantilever that matters. To be more precise, the electrostatic actuation of our first eigenmode increased the noise a little bit so we started to decrease the amplitude of this excitation to almost get to the expected thermal noise level. 
   
• Q4 : Is there a reason why bulk conductive tips, such as Rocky Mountain Nanotechnology^TM, cannot be used?
• A4 : For tips without a coating layer, if you measure short range forces, your tip shape doesn’t matter that much. But for long range forces, such as magnetic force, your tip shape matters a lot, for this reason we use high-aspect ratio tips. We coat them from one side because we want to have a reliable magnetic tip where we can flip the magnetization of the tip. Especially in 3D magnetism, which can exhibit all sorts of transitions. We want to have a macroscopic dipole, not microscopic, that we can flip and not get random transition from one state to the next. That’s why we use coated sharp tips to get more stable, well defined dipole moment. And the second reason is simply, well, the resolution. That’s why we don’t like bulk tips, but for electrostatic measurements such as SSRM (Scanning Spreading Resistance) or PFM (Piezoresponse Force) when you go into contact then you need them. We could also be more inventive such as Noncontact PFM and we already tried, which also works and that could be the future for those techniques, to stay out of contact and use sharp tips all the time.
   
• Q5 : My question is related to bimodal AFM because if you have large amplitude at lower frequency and small amplitude at higher frequency, your tip is still far away most of the time with small oscillations. Maybe my question can be directed toward Zurich Instruments: could you trigger and just measure at the lower turnaround point of the slow frequency (first eigenmode)
• A5 : The first mode is not affected by the second one, because the frequency are incommensurate. Nature helps us here because the second mode is 6.28 times higher. There is this old paper from Shigeki Kawai, Alexis Baratoff, Thilo Glatzel, which addresses this, and it’s called time averaged SPM. If you do a frequency shift vs distance curve in small oscillation mode and small & large oscillations together, the shape of the curve is very similar, only the excursion is smaller so that the total recorded frequency shift is smaller. The mathematics is tricky, but it shows that you really get the same signal.  Some handwaving arguments is that the signal is mostly measured at the lower end of the cycle because the tip changes direction and is much slower than when it moves to the opposite end. It of course also spends more time at the opposite end but then there are no signal to measure (far away). So basically, you average your signal when the tip is closest to the surface and that’s why such signal corresponds so well to the signal measured with one mode only. Maybe a gated measurement only at the lower end could corroborate that too and we should try that with Zurich Instruments.
   

Cryogenic AFM instrumentation and automation with Python API

Background: Prof. Yoichi Miyahara starts with single electron sensitivity discussion in EFM, followed by the description of his new LT-AFM in a closed loop dilution refrigerator. The second part of his talk discusses cavity optomechanical effects and how this affects the effective Q-factor. Finally, the MFLI integration with GXSM via python API and automation of Frequency Modulation KPFM mode is presented.

Q&A session:

• Q1: My question is related to the optical force, you said that the radiation pressure has very small contribution to the overall force, like 4%. Could you use this to cool your cantilever?
• A1: We use platinum coated cantilever, which absorbs light and therefore increase the temperature. This is what limit us at this stage for cantilever cooling. Some dielectric coating would be needed in this case. But the ratio Q/T would anyway remain the same

• Q2: How are you satisfied with your 2 laser systems? usually people excites at the base and measures at the tip but you excite and measure at the position of your optical fiber where the tip is. How does it work, it this really reliable?
• A2: At low temperature, it works really well, we surprisingly need only a very small amount of force so there is no issue with the excitation if you apply the force at the tip of the cantilever, while we take advantage of this spurious free resonance response.

• Q3: Would this work with low Q resonator as well?
• A3: Depending on the absorption layer and optical power we can excite stiff cantilever as well with low Q, such as in liquid. 

Acknowledgement

We are indebted to the NC-AFM organizing committee, namely Peter Grutter, Omur Dagdeviren and Catherine Boisvert for allowing us to organize such an event within the context of the conference. Thanks also to Jorge Medina, from our US office to join force during this event and at the conference as well as the local McGill AudioVisual Team, namely Stewart McCombie and Jonathan Roy for a smooth broadcasting and recording experience.