Developing a Quantitative Nanoscale Characterization Technique

Interview conducted by Will SoutterOct 17 2013

In this interview, Dr. Chanmin Su, Director of Technology for Bruker Nano Surfaces' AFM business, talks about how Bruker's philosophy of innovation applies to his part of the business, and tells AZoNano about the development process for Bruker's groundbreaking PeakForce Tapping™ AFM mode.

The suite of PeakForce Tapping modes is being rapidly adopted by the community, and is poised to displace the venerable TappingMode™, which in fact, was another invention from the Bruker lineage. Before we get into the details of PeakForce Tapping, can you tell me a little bit about Bruker AFM and its philosophy of innovation?

Certainly. I think all major technology companies are fundamentally driven by innovation, and Bruker is no different. Our level of investment in research and development is amongst the largest for technology companies. We have also formed a collaborative network with the leading research groups and institutions in the world.

For example, we have a long-standing relationship with Paul Hansma, Cal Quate, and Christopher Gerber. All three of them are basically founding fathers of AFM, and we are working with them constantly.

This network serves two main purposes: firstly to keep abreast of the latest developments at the cutting edge of the technology itself, and secondly to recruit talent from the institutions we work with, to maintain an environment of innovation.

As a result of our R&D efforts, we have generated the largest patent portfolio in the AFM field.  We have also released more new products than any of our competitors – with multiple major new products launched every year.

Experimental data of force curves for a cantilever operated in PeakForce Tapping. The lever is driven by a sinusoidal wave and the curves are displayed as force versus time and force versus distance. (Images from Bruker Application Note #133.)

What led to Bruker’s initiative to develop PeakForce Tapping?

The idea actually started much earlier than when we began developing the product – the real starting point was in 2004, when the National Nanotechnology Initiative offered grants for solving a series of nanotechnology challenges. One of these challenges was quantitatively characterizing properties of materials at the nanoscale. We took this challenge on at Bruker, organized a team to brainstorm possibilities, and we eventually won a $13 million grant from NIST for quantitative nanomechanical property mapping, in conjunction with Dow Chemical – this significant investment in PeakForce Tapping, is one of the things that enables its sophisticated implementation and truly sets it apart from other technologies.

The grant was based on a three year project to develop both a fundamental understanding of the necessary technology, and also to show commercialization potential for the quantitative tool. During the development process, we worked with leading groups, including Karl Kreig’s group and a scientist at Stanford named Dr. Ozgur Sahin. Ozgur originally came up with the idea that became HarmoniX, which we commercialized in 2008. HarmoniX, as a precursor to PeakForce Tapping, was a way to map quantitative mechanical properties at the nanoscale, but something was missing..

AFM is really all about force control – hence the “force” in atomic force microscopy. The holy grail for AFM is to be able to control and acquire both the instantaneous interaction force, as well as all the interaction data between the tip and the sample. HarmoniX , likeTapping Mode, does not actually achieve this, so there was a fundamental drive for us to push these technologies further, and that is what led to the development of PeakForce Tapping.

We conceived of PeakForce Tapping towards the end of 2008, and brought it to market quite quickly – we productized it in 2009, with the first AFMs featuring it as a standard component being released in 2010.

What are the basic principles behind your PeakForce Tapping modes?

If I were to describe in a single sentence what PeakForce Tapping is about, I would say it’s about simplicity. Thinking about the AFM at its most basic level, you are using a stylus to touch a sample. If you think only about a single “touch” point (approach, “touch”, and retract) you generate a force map, which has a sort of pulse shape similar to an electrocardiogram – except the peak of each pulse represents the stylus contacting the surface.

As PeakForce Tapping maps the surface, we use feedback to keep that peak force constant, down to 10 or 20 pN – in both air and fluid. Because of this simplicity and control, we can keep the tip as sharp as a nanometer, which gives us atomic resolution.

30μm scan of a Teflon membrane in PeakForce (left) and regular TappingMode (right). Artifacts visible in TappingMode operation are not present in the PeakForce Tapping data. (Images from Bruker Application Note #133.)

What are the practical benefits of PeakForce Tapping over conventional tapping modes?

I would say the major benefit of PeakForce Tapping over TappingMode is that the AFM is finally performing imaging by direct control of interaction force – and more importantly, control of the instantaneous interaction force. Tapping measures an indirect average force and this is a critical distinction. In TappingMode, we can control the amplitude, or the phase – neither of which are a direct reflection of the force. They include a whole host of complications: the cantilever dynamics, the environment (fluid/ambient), as well as the tip-surface interaction and adhesion. Therefore, it is very difficult to judge the right amplitude/phase for a specific measurement – it takes time and expertise.

The second most important difference is that AFM operates on a feedback loop. SEM, TEM, and optical microscopy are all open-loop – you shine light or an electron beam at a sample, receive the signal, and create the map directly. With AFM, because it is a closed-loop system, it is error-driven – error is corrected using the closed feedback loop.

This is very much like when you drive. If you are driving a car on a winding road, you use information from your eye (the sensor) to correct any deviations from the road using your hands (the actuator). This is a good example of how a feedback loop works.

So in AFM, the way we generate the error and track the error are crucial parts of the design of the system. We track the error using a PID loop, which has a linear response. In TappingMode, however, where we are controlling amplitude or phase, the error generated is extremely non-linear – so you have non-linear error generation with linear tracking, which obviously don’t match.

So it takes years of experience to match the linear feedback loop to the non-linear error to remove artifacts, and this is what fundamentally makes complete automation of the AFM’s feedback impossible. We do have some automation in our systems for simple silicon semiconductors, but with anything more complicated than that it becomes very difficult to optimize the control parameters and generate what I call “expert” data with TappingMode.

This makes the traditional AFM a very elite instrument. Only a really experienced user, with a degree in physics or engineering, who has worked on different samples for years, and knows AFM very well, can arrive at a point where they have an artifact-free image, and are not confused by some complication with the feedback or control.

So PeakForce Tapping brought an important change to the field, because it allows a novice user to generate “expert” data.

These two points, direct force control, and being able to feedback on that linear response, really summarize the biggest benefits of PeakForce Tapping – allowing new users to match that quality of image, even on a complicated sample like a multi-component polymer, and at the same time get rich high-resolution quantitative nanomechanical data.

So how is the feedback loop in PeakForce Tapping different – and how does this make quality images easier to obtain?

It’s all based on the fact that we are controlling the peak force, instead of the amplitude and phase. The error in the peak force grows linearly with the interaction, because it is insensitive to the sample or environment – which fits very well with the linear PID loop – this makes the error correction much easier, and as a result the system is much more stable.

Another benefit is that when you change the set point, the force response is very direct. In TappingMode, when you increase or decrease the set point, the force goes through a very complicated curve before it restabilizes. With PeakForce Tapping, the change to the new set point is direct. This, plus the linear feedback loop, allows us to build this control into the system. We call this feature ScanAsyst^® and it is a self-optimizing control system for the AFM – basically single-button AFM imaging.

Finally, the accuracy and repeatability that this mode enables has been very popular with our customers. Whenever I go out to see our industrial customers, they are mostly using PeakForce Tapping because it allows their technicians to get repeatable, artifact-free data, even from a new sample, without having to start a whole new research project to get the methods right.

"ScanAsyst is a self-optimizing control system for the AFM - basically single-button AFM imaging."

What properties do the other PeakForce modes (TUNA™, QNM™, KPFM™) allow access to, that can't be measured using standard contact or tapping modes?

There is a lot of information that we can take from the force curve in PeakForce Tapping. If we go back to the electrocardiogram analogy - a doctor can look at an ECG trace and judge, for example, whether the heartbeat is normal or not.

We can do the same with the force curve in PeakForce Tapping - we can go much further than just mapping the surface topography. From that single force curve, we can derive elastic modulus, adhesion and dissipation, and other parameters that are unique to AFM.

These are all properties that are inseparable when you use TappingMode, because they are non-determistically combined into the phase, like a linear equation with three or four variables. In PeakForce Tapping, what we are doing automatically is separating out the topographical map from the adhesion and elastic properties - providing solutions for each of those variables individually.

That is just for the mechanical map - what we have also done is added conductivity mapping into PeakForce Tapping to make TUNA (Tunneling AFM). Conductivity mapping is nothing new - it is easy to do with contact mode. However, contact mode cannot map soft or sensitive samples without permanently changing the sample and potentially spoiling the data.

I should note that conductivity mapping doesn't work with TappingMode either, because there is no way to know exactly when the AFM tip is touching the sample. With PeakForce Tapping, we can use the force curve to know the exact period when the tip is in contact with the surface, and extract the conductivity data from just that moment. So that's how we were able to develop electrical mapping for soft or heterogeneous samples with TUNA.

There are also benefits in integrating PeakForce Tapping with Kelvin-potential mapping, another established technique that measures force gradient as well as the force itself. For both parameters, the sensitivity of the measurement is greater the softer the spring constant of the cantilever. TappingMode therefore creates an inherent limitation on the sensitivity because the cantilevers used in Tapping Mode are very stiff - around 40 N/m.

PeakForce Tapping removes that limitation because it can use 0.3 N/m cantilevers - making the contact more than a hundred times softer, and therefore allowing much more sensitive Kelvin potential measurement. You can also use a much sharper tip with PeakForce Tapping, which increases the resolution of the map as well.

Cycle of heating and cooling of polymer blend of syndiotactic polypropylene and polyethylene oxide. Images a–f show height of the surface during the process, while g–i show the modulus of frames d–f respectively. Scan size 5um. (Images from Bruker Application Note #128.)

PeakForce QNM is one of the most powerful and revolutionary modes on the AFM market - what new applications for AFM has this mode enabled?

Yes - PeakForce QNM is really the accomplishment of the original challenge from the NNI, to measure material properties quantitatively at the nanometer scale.

So there are many ways this capability can be applied. One example is in the chemical industry, for companies like Dow Chemical, Dupont, 3M - all of whom make heterogeneous polymer materials of some sort.

3M are mainly focused on multilayer polymers, so they would want to know the mechanical map of the cross-section of those layers, where each layer is less than 100 nm thick. Potato chip packets are a good example of these multilayers - they are actually quite a complex material! There is no other tool besides PeakForce Tapping that can mechanically map those layers at the sub-nanometer resolution.

Similarly in impact plastics for cars and civil engineering, Dow Chemical makes impact plastics that are actually composites of at least five components, and many of those components are in domains smaller than 100 nm. For example, a composite might contain a soft polymer to absorb shock, and a hard component to enhance strength. In these kinds of materials, it is crucial to be able to characterize the load transfer between components, at the nanoscale. PeakForce Tapping can do this, quantitatively and with good repeatability.

Another very important application is characterizing defects in semiconductors and data storage. A particular defect could be a contamination, or it could be an internally grown defect of the same substance as the device. In process control, knowing where defects come from is crucial to diagnosing issues.

This is getting harder to do with old techniques, since semiconductor architectures are getting smaller and smaller - 14 nm modes have been announced, and manufacturers are looking towards 10 nm modes and below. There could be really crucial defects that can no longer be seen with optical microscopy, and eventually even with SEM, especially with organic defects.

With PeakForce Tapping AFM, you can map out those defects, and provide quantitative mechanical information. This leads to a new way to characterize defects - you can tell whether it is a polymer left over from the photomask, or if it’s metallic, or an organic contaminant. It isn't directly a chemical identification, but it gives you good knowledge to work out what the defect is if you have some a-priori information.

AFMs like the BioScope Catalyst are becoming more and more popular as a tool for life science research - what do Bruker’s PeakForce Tapping modes offer for biological analysis that has helped make AFM so popular in that field?

I think PeakForceTapping is attractive to biologists for one simple reason: it's simple. Biology researchers don't have to take a physics lesson to understand cantilever oscillation in fluids, and there's no tuning and optimizing to do. It really makes AFM easy to understand conceptually as well - it's just like touching a surface with your finger. You can tell if it’s rough or smooth, soft or hard.

This is really important for biologists, because they are using AFM as a tool to study biological materials, not to study physics - the bio researchers don't need any additional knowledge base to make it work for their research - even in fluid.

Another important factor is that PeakForce Tapping gives you a mechanical map, not just a height map - the height is really irrelevant when you are imaging something like a cell, because cells can change shape arbitrarily. What you really need to know are the cell’s mechanical properties.

In the future, what I’d really like to see users doing is creating a full molecular-scale map of a live cell surface using PeakForce Tapping - right down to identifying specific protein sites. That is where the biological AFM research is going, and I anticipate that happening in the next few years.

Typical application of PeakForce QNM imaging on living plant cells. (A) Projection from a confocal stack of an Arabidopsis Thaliana shoot apical meristem. Membranes were labeled with FM4-64. (B and C) PeakForce QNM images (top: 3D-height, topography only; bottom: 3D-height with DMT modulus skin). The DMT modulus channel clearly indicated that the cell edges (anticlinal cell walls) were significantly stiffer than the rest of the cell. Circled areas show regions where the modulus and optical maps reveal the presence of anticlinal cell walls that are not detected when using topography alone. (Images from Bruker Application Note #141.)

So it’s clear that the ease-of-use that the PeakForce Tapping modes bring to AFM are incredibly valuable for a lot of your users. Do any of the expert users that you’ve mentioned now feel like this has almost devalued their expertise to some extent?

Perhaps originally, now though, two years on, there are very few experts we talk to who still think that way. Even the most experienced scientists use PeakForce Tapping because it is reliable and judgement free - and produces data just as good as they can manually.

We’ve also done some tests comparing expertly tuned TappingMode data with one-button PeakForce Tapping data generated by a novice, and in a lot of cases the PeakForce Tapping technology actually produced a better image than the expert.

Do you have any plans that you can tell us about to add more property- or application-specific modes to the PeakForce Tapping range?

There a couple of areas I can think of that have great promise. One is the integration of optical signals, where you can map physical and chemical properties simultaneously. This can be achieved by acquiring a near-field optical signal on the same probe, so that the optical data adds an extra dimension to the AFM map. This would be a great piece of integration, to give you a multi-dimensional nanoscale map of a complex system in one pass.

There is a lot of work going on already with the integration of optical spectroscopy with AFM, and I think that is going to continue as a great way to enhance the mechanical map with chemical data.

The other way to get that chemical data onto the map would be to convert PeakForce Tapping into a pure chemical mapping tool.

How is this possible? Well, we could chemically functionalize the probe, so that the probe has some molecular recognition ability. You can do this to some extent already, but there is no way to create a full chemical map. PeakForce Tapping does have the capability to let us try this out in the future, with a variety of functionalizied tips - effectively making the AFM probe a biochemical tool, rather than a mechanical tool.

This would be a great addition for many of the applications we have talked about - mapping out molecules on a live cell surface, mapping any protein, or crystal, or any number of things. I see that as a great direction for the future, but there is a lot of development to be done, which we expect to be doing ourselves, in collaboration with, and alongside our scientific users, to achieve the necessary chemical specificity through surface functionalization.

About Chanmin Su

Chanmin Su is the inventor of 18 patents used in AFM systems, including PeakForce Tapping. He received his Ph.D in solid-state physics from the Chinese Academy of Sciences in 1988, and pursued post-doctoral research at KFA Forschunszentrum in Germany. In 1991, he worked for the University of Maryland as a research associate and then as an assistant research professor.

Moving to industrial research in 1998, Chanmin took the position of principal scientist at Raytheon Systems, developing a MEMS barometer for millimeter infrared detector arrays. Chanmin Su subsequently joined Veeco Instruments in 2000 as a senior staff scientist. He is currently serving as Director of Technology in AFM business at Bruker Nano Surfaces, overseeing technology development.

Chanmin has published over 60 papers and two book chapters on the topics of the mechanical properties of bulk/thin film materials and AFM instrumentation, and has been the co-organizer of several international conferences on scanning probe microscopy, as well as a panelist for several national and international reviews on future technologies.