This blog post accompanies the webinar, where we hosted Prof. Peter Grutter from McGill University in Canada. In this webinar, Peter explained some inner mechanisms of chemical reactions in molecules and charge dynamics in solids, that determine materials' most fundamental properties. To achieve this, he developed in the Nanoscience & SPM group techniques that combine nanometer spatial and femtosecond temporal sensitivity, bridging the gap between the AFM and laser communities.
After the talk, we shared practical tips from the Zurich Instruments' toolbox, which were used in several of the experiments presented in Peter's talk. These tools included PLL/PID for NC-AFM, the ability to control several instruments from the same LabOne interface with multi-device synchronization (MDS) and the boxcar averager or arbitrary waveform generator for time-resolved SPM.
Thanks again to everyone who joined us during the live event and actively participated with many great questions! The answers to most of these questions are available at the end of each section.
Measuring electrostatic forces by AFM
Highlighted topics addressed during Peter’s talk:
- Beware: C(V) dependance in semiconductors exhibits non-parabolic response in KPFM (Kelvin Probe Force Microscopy) -> 12’34’’
- Using the AFM tip as a local nano-electrometer down to single-electron sensitivity -> 15’07’’
- Imaging coupled quantum dots: stability diagram with a scanning tip -> 18’36’’
- Electrostatic Force Spectroscopy on single molecules -> 24’26’’
- The limit of time resolution with AFM -> 29’24’’
- Pump-probe KPFM -> 30’30’’
- Nonlinear optical signal measured via AFM -> 35’22
Q&A related to the first section of the webinar
Here is a summary of the main questions asked during the webinar. Some were answered live.
- What is the fundamental difference between an all-optical pump-probe and a force sensing time-resolved spectroscopy?
In all optical pump-probe, you detect propagating waves (high frequency), in force detecting part, you measure DC components. It’s the same physics at different frequencies. Mechanically based detection has a small probing volume with the tip which makes the phase matching condition to do non-linear optics much smaller. Effectively, the two optical beams are always phase matched. The drawback is to watch out not to heat the tip or cantilever too much as this would act as a disturbance. And ultimately, you can get down to atomic-scale structures with optical time resolution, which is not possible in all optical detection.
- Which KPFM technique would you consider best suited for the type of semiconductor sample that you showed that exhibit bend bending effect?
Dissipation KPFM is relevant when there are no dissipation processes other than electrostatics to be measured. It is therefore not well suited for single electron tunneling that exhibits back-action effect. For semiconductors, it is well suited especially when measuring already in vacuum with non-contact AFM and at a higher modulation frequency than standard FM-KPFM (at the same frequency of the mechanical resonance or second harmonic).
- How do you determine the thickness of tunnel barrier layer in the sample?
For thiols it is known as a f(chainlength). Didn’t show it, but we have also engineered NaCl barriers - we just empirically determined it. InP (the InAs Quantum Dot substrate): was a coincidence, and once we knew, you can control the growth, that’s what we did. :)
- What do we have to consider when performing pump-probe measurements in air instead of UHV?
Fundamentally, there are no issues in doing pump-probe experiment in air. In ambient conditions, time resolution is limited by the thermal limit. Q-factor is lower in air compared to vacuum, which leads to lower force sensitivity, and you need to average longer. This can generate problem in term of drift stability, such as how much tip-sample distance changes, how much the lateral tip is drifting. This is better in vacuum and best at low temperature. But if you don’t need to achieve femto-second time resolution, you can do the same experiment in air. We actually do it sometime in air because it’s easier from a sample preparation point of view.
- Would the incident angle of the pulsed laser impact the measurement? what is the typical incidence angle of the pulsed laser?
The incident angle is very flat, few degrees. The polarization has a huge effect. Ideally, we would like the beam to arrive parallel to the surface. This impact the effective polarization vectors on the sample. If the angle is too large, you can also heat the cantilever.
- Taking about the effect from the thermal noise, it would be quite straightforward to go to the cryogenic temperatures. What would be the typical thermal load from the pulsed laser?
Low enough - power levels are not that high and you can see when you ‘heat’ the cantilever. Typically, we heat the cantilever by a few degrees - which we can measure as a frequency shift that stops changing after ~20’.
- In an electrical pump-probe technique, is the electrical pump via the cantilever? If so, then is cantilever dynamics not a limitation?
We typically pump the sample. The reason is because dispersion in the cantilever etc. changes the pulses shape of fast pulses.
- Does the tip has really no effect, even if it is an oxidized silicon tip?
Great question: we use only metal coated tips ! SiO tips have a great optical response…
- Can you exploit the non-linear optical effect of the tip in close vicinity to the surface to improve resolution and relieve the constraint of the sample itself having non-linear optical properties?
Possibly - though we tend to stay ‘far’ from the surface as getting close (to generate non-linear effects) also leads to a whole bunch of other challenges, such as heating. Our signal does not rely on a rectification or non-linearity between tip-sample.
Nevertheless, we are not yet at the stage where we quantitatively understand the magnitude of our signal. Working on that :)
How smart instruments help build complex experiments
Highlighted topics addressed during Romain’s tutorials:
- Why NC-AFM is quantitative using orthogonal analysis, maximum amplification, and constant transfer function? -> 53’24’’
- How to make sure your NC-AFM set-up is optimized and artefact-free? -> 55’47’’
- How to grow the experiments with your need using Multi-Device Synchronization -> 1:05’29’’
- 3 approaches to Time-Resolved SPM -> 1:10’31’’
- 2 Boxcar integrators as a sorting tool for on/off illumination cycles -> 1:13’54’’
Q&A related to the first section of the webinar
- What is the smallest boxcar window you can make?
The smallest boxcar window depends on the UHFLI ADC sampling speed, which is 1.8GSa/s. You can therefore expect boxcar window as low as a nano-second. This doesn’t prevent you from measuring femto-second events that are averaged over many repetitions rate and that generate a detectable change within that time window.
- Is the AWG option available on MFLI or HF2LI as well?
The AWG is only available on the UHFLI version or as a stand-alone version. But stay tuned, we will have more instruments featuring this option in the future.
- Is it possible to track two eigenmodes of the cantilever at the same time with PLL setup with one lockin only?
Since the 2 eigenmodes would be excited and detected all mechanically, therefore using a single output and a single input, it can be all be done with a single MFLI with the MF-MD multidemodulator and MF-PID option. As a matter of fact, you can even output the linear superposition of all 4 frequencies of 4 eigenmodes on the same signal output and track them with the equivalent of 4 PLL. Having 2 MFLI working as one via MDS makes sense when you need let say 2 outputs, one as mechanical drive and one as electrostatic drive.
- Are the PLL automatically synchronized between 2 MFLI when using MDS?
MDS synchronizes the 10MHz reference clock and time-step across instruments. Arbitrary frequencies such as the one locked by a PLL would need to be synchronized as an external reference signal to the second MFLI. This uses one demodulator but doesn’t require the MF-PID option on that instrument since ExtRef is standard for all instruments.