Non-Contact Mode AFM and Dimensional Nanometrology for Inline Manufacturing

Interview conducted by Will SoutterFeb 25 2014

In this interview, Dr. Sang-il Park, CEO and chairman of Park Systems, talks to AZoNano about his company’s current business position in industrial manufacturing, and the benefits of Park instruments for semiconductor failure analysis laboratories.

Park Systems is amongst the oldest AFM technology companies, and you were involved in the field from the very beginning. Can you tell us how you became involved in the AFM world, and how Park Systems started?

I was very fortunate to be at Stanford at that very early stage, just when the AFM was being invented. After studying physics at Seoul National University in the early 1980s, I moved to the U.S. to further my career, like many other Korean scientists at the time, choosing to do a PhD at Stanford University.

My stroke of luck was that the PhD supervisor was Calvin Quate, who was developing the world's first AFM at the time. I contributed quite a bit to Quate's group, and saw first-hand how the AFM and then other exciting extensions to the technique like magnetic-force microscopy were brought in.

AFM was invented in Prof. Quate’s Lab at Stanford Univ. in 1986, where Sang-il Park was a graduate student in the group. Sang-il Park founded Park Scientific Instruments in 1988, and commercialized AFM, the first in the world.

I realized that the AFM was a not just a research tool, but also had great commercial potential. I took the ambitious step of setting up a firm called Park Scientific Instruments (PSI) in 1988, serving as chairman and chief executive. To begin with, I ran the company out of my garage – however, we soon moved to a larger commercial location, and the company enjoyed considerable success thanks to our first-mover advantage.

PSI broke even in its first year, and was profitable in the second - after 9 years, in 1997, I sold the business to Thermo Spectra and moved back to Korea. There, I founded the second company, PSIA, developing next-generation AFMs in partnership with the Thermo Spectra. PSIA became a fully independent entity in 2001, and was renamed Park Systems in 2007.

AFM is becoming more and more popular in the industrial sector, as well as in research, and Park Systems has announced great results in this area – particularly in the data storage market. What are the reasons behind this shift in the AFM industry, and Park’s success in these expanding markets?

Our recent successes, which include capturing 90 percent of the hard disc industry market, were accomplished by simply outperforming our competition. The conventional AFM technology still employed by our major competitors has been riddled with problems and limitations related to fundamental aspects of instrument design.

To remedy these issues, Park created a new architecture, and perfected the non-contact mode, which has drastically increased accuracy and usability while bringing down the total cost of ownership of AFM tools. The performance of our non-contact technology has made our instruments very attractive to hard disk manufacturers.

In addition, Park Systems has achieved a growing market share in industrial markets through a focused effort on customer needs and performance requirements, resulting in a more desirable AFM product. Park’s customer service is offered virtually round the clock from both Korea and from the U.S. offices, utilizing both remote control and on-site support, adding to customer satisfaction and loyalty in the long run.

We are constantly developing and enhancing our products to meet the ever more stringent industry requirements, and continue to invest heavily in R&D to provide the most advanced AFM for failure analysis (FA) and quality assurance (QA). In this highly competitive market, where a transition from one vendor to another means eventually replacing all the tools, and with each costing millions of dollars, our growing market share is a testament to Park’s great customer service, and the strength of our technology.

If possible, could you share the details of your company’s successes in the industrial market? What new capabilities have been incorporated into your AFMs and made the difference for your industrial customers?

Our initial success came from the hard disk drive (HDD) industry, specifically slider manufacturing, and now it’s expanding to other industries such as semiconductors. Process engineers in slider manufacturing in the HDD industry use AFM for Pole Tip Recession (PTR) measurements, a critical process for monitoring hard disk drive failures.

We have developed a fully automated AFM, the PTR Series, for PTR measurements for the HDD sliders with nanoscale accuracy and repeatability, thus improving the overall production yield.

Park Systems' industrial AFM provides accurate and reliable Pole Tip Recession (PTR) measurements in full automation.

Traditionally, PTR measurements by AFM typically required multiple scans - macroscale reference scans along with higher-resolution scans of smaller regions of interest. This multi-scan process takes time, and limits throughput in AFM measurements. Our crosstalk-eliminated scan system allows for truly flat scans, effectively eliminating the multi-scan process.

In addition, True Non-Contact™ mode preserves the sharpness of the AFM tip for prolonged high resolution imaging and much longer tip life. As a result, our AFMs can generate accurate images of highly detailed regions of interest within larger macrostructures, without any need for reference scans to correct various scanner artefacts.

After building our reputation in slider manufacturing, we were invited into other sectors of the hard disk industry: media and wafer fabrication. The task of identifying nanoscale defects is a very time-consuming process for engineers working with media and flat substrates.

We developed a capability, automatic defect review (Park ADR), which speeds up and improves the way defects in substrates and media are identified, scanned, and analyzed. This works by taking a defect map from an optical inspection tool, and imaging the individual defects in high resolution with the AFM. This automated process allows for very high throughput in defect imaging.

Using the defect location map provided from an optical inspection tool, Park ADR automatically goes to the defect location and images the defects.

Our success in the hard disk industry resulted in attention from the semiconductor industry as well. Here, we faced another problem that required a new approach in AFM design: sidewall roughness. It is well known that due to the shape of the tip probes, AFMs struggle to measure structures with vertical walls. To cope with this, cantilevers with flared tips have been used in the past. These work well for simple overhang structures, but are not capable of producing a detailed image of the sidewall. As sidewall roughness measurements are a crucial part of semiconductor quality analysis, this is a major limitation.

To overcome this, we were able to develop a new AFM technology that allows nanoscale measurements with unprecedented detail on both metal and photoresist sidewall structures.

One of the unique features of Park's AFMs is the implementation of an independent Z-scanner that is decoupled from the XY scanner. The Z-scanner can be tilted in order to gain access to the sidewalls. Hence, the combination of three scans, each one performed at a predefined angle (typically 0º, qº, and -qº) allows for the reconstruction of the full 3D pattern. Unlike flared tips, the probes that are utilized on the new 3D AFM from Park maintain the high aspect ratio and sharp tip radius (<10 nm) providing accurate dimensions along with sidewall roughness parameters.

Decoupled XY and Z scnning system; 3D FM using tilting Z scanner.

Imaging sidewall roughness with Park 3D AFM.

In addition, by utilizing the True Non-Contact mode with these probes, even soft and sticky samples, such as photoresist, are reliably measured without any deformation or damage. Non-contact 3D-AFM can ensure accurate measurements of the critical dimensions of top, middle, and bottom photoresist lines while simultaneously performing roughness measurements along their sidewalls.

Can you tell us more about how True Non-Contact mode works? How were you able to achieve non-contact scanning in ambient conditions?

We have never kept the details of ambient non-contact mode AFM a secret. The main challenge of enabling non-contact scanning at ambient pressure is well known and widely published: fast Z servo feedback. Our innovations in AFM scanner design and feedback control over the years have been focused on attaining higher Z-servo speed.

The Wickramasinghe group at IBM developed the first non-contact mode AFM in 1987, one year after the development of the contact AFM.In non-contact mode, a piezoelectric modulator is used to vibrate the cantilever near its intrinsic resonant frequency (f₀), usually between 100 kHz and 400 kHz, with amplitude of a few nanometers.The modulated feedback operates in the attractive interaction regimes of the van der Waals force.

The challenge with non-contact mode lies in maintaining the tip within the regime of net attractive force, which is very small as we can see from amplitude vs. distance plots.

Amplitude vs. distance plot of a tip oscillating with small free air amplitude.

This is why rapid Z-servo feedback is so crucial – if the Z-axis performance is too slow, the scanner falls into bi-stable mode hopping, where the tip will oscillate between the repulsive and attractive force regimes (contact and non-contact modes).

Most AFM vendors, lacking the Z scan actuators with the feedback control necessary to stay in the attractive force regime, elect to operate their systems in the repulsive force regime, in “Tapping Mode”. This means that their tips periodically come into contact with the surface of the sample, wearing down the AFM tips, and potentially damaging the sample surface.

What led you to develop Non-Contact mode originally? Was there a specific application you had in mind that would benefit from it?

The initial motivation was the need for more accurate dimensional nanometrology for inline manufacturing control, especially in the hard disk and semiconductor industries.

While most AFMs generate impressive three-dimensional images, many are not designed for advanced metrology. They function well as a qualitative imaging tool, but lack the accuracy required to monitor critical process parameters for manufacturing control.

For example, the read heads of hard drives fly over the disk surface in such close proximity, that if PTR is too high by as little as the height of an atom (about 3 angstroms), the electrical performance yield drops by 10%. If the PTR is lower by about one atomic dimension, 5% of the hard drives produced would crash and fail.

Proper monitoring of the PTR value requires a measurement accuracy of 0.1 nm, referenced against a surface which is 20µm away. This is equivalent to repeatedly measuring 1mm step heights from 20m away, down to the accuracy of a human hair (about 0.1mm).

Pole Tip Recession measurements are fully automated with the upcoming Park NX-PTR, providing a high throughput capability, for both at the carrier row bar level and at the slider level.

Proper monitoring of the PTR value requires a measurement accuracy of 0.1 nm, referenced against a surface which is 20µm away. This is equivalent to repeatedly measuring 1mm step heights from 20m away, down to the accuracy of a human hair (about 0.1mm).

Due to the design of traditional AFMs, they display intrinsic bowing in the range of 50nm over the 20µm scan. This can be corrected for by subtracting a best-fit second- or third-order plane, but this makes the measurement of PTR (typically around 1 nm) more of an art than a science for most instruments.

Park AFMs are designed and optimized for industrial metrology, providing a highly orthogonal and flat scan with adequate repeatability and accuracy for precision nanometrology. Also, in True Non-Contact™ mode, tip-sample interaction is very weak, which minimizes wear on both the tip and the sample during scanning.

Your recent product, the Park NX20, has been specifically tailored to industrial applications. Is there a particular type of analysis that this new product excels at?

The electronics industry is increasingly pushing for ultra-flat substrates, to meet the demands of the ever-shrinking dimensions of devices. This means that FA and QA engineers are facing the daunting challenge of controlling sub-angstrom variations in surface roughness or film thickness.

This puts stringent requirements on performance of AFM scanners. In order to meet today’s process control requirements, scanner artifacts must be kept to an absolute minimum. The out-of-plane-motion (OPM) must remain below 1nm regardless of scan sizes, scan offsets, and scan rates. In addition to the OPM, the XY orthogonality has become a critical factor, too, and the deviation from perfect 90 degrees needs to be lower than 0.2 degrees.

The XY scanner consists of symmetrical 2-dimensional flexure and high-force piezoelectric stacks provides high orthogonal movement with minimal out-of-plane motion, typically less than 1nm, as well as high responsiveness essential for precise sample scanning in the nanometer scale.

The NX20 is designed explicitly to deal with these challenges. It is equipped with a closed-loop XY scanner with Dual Servo system for increased accuracy. Each axis of the scanner has two position sensors – the second sensor provides correction for any orthogonality errors in the first, maintaining high orthogonality even with the largest scan ranges and sample sizes.

Dual Servo System is a two symmetric, low-noise position sensors are incorporated on each axis of the XY scanner to retain high scan orthogonality for the largest scan ranges and sample sizes.

Non-contact mode has also got even better on the NX20. We have quadrupled the Z-servo bandwidth compared to its predecessor, and this helps reduce tip-surface contact even further, to maintain a sharp tip, and therefore accurate and repeatable roughness measurement, for much longer

Also, the NX20 is equipped with the most effective low noise Z detectors in the field, with a noise of .02 nm over large bandwidth. This produces highly accurate sample topography, no edge overshoot and no need for calibration. In conventional AFM scanning, the topography signal is plotted from the calibrated voltage applied to the Z-scanner. However, errors known as piezo creep occurs due to the intrinsic materials property of a piezoelectric actuator that drives the Z scanner. The only way to correct this artifact is to use an independent position sensor that directly measures the topography.

When the topography signal is the applied voltage to the Z-scanner, errors known as edge overshoot on leading and trailing edges often occur. With its industry-leading low-noise (0.2 angstrom) Z-position detector, the Park NX20 overcomes the piezo creep errors and generates topography without edge overshoots.

In addition to these improvements, the NX20 is also equipped with an extremely friendly user interface, and extensive automation capabilities. Park’s automation control software can accurately collect data, perform pattern recognition, and do analysis using its onboard Cognex board and optics module, and export results with almost no user input so you have more time to do your own research.

Park NX20 Atomic Force Microscope

You recently announced QuickStep Scanning Capacitance Microscopy (SCM) and PinPoint Conductive AFM for the failure analysis of semiconductor devices. The two new modes give Park a strong position as the failure analysis solution provider. How do the two new modes differ from the conventional SCM or conductive AFM?

In short, QuickStep SCM is several times faster than conventional SCM, without sacrificing any measurement accuracy. PinPoint Conductive AFM provides frictionless, on-spot conductive measurement that is highly accurate, and repeatable with no or little tip wear, unlike the conventional method.

In SCM, one needs to get a good signal-to-noise ratio (SNR) for high sensitivity measurements. The conventional SCM methods attained the necessary SNR by very slow, continuous movement of the XY scanner. In QuickStep SCM, the XY scanner moves quickly to each measurement pixel point, records the data while dwelling on the point, then moves rapidly to the next measurement point. This speeds up the scan rate significantly while maintaining the high accuracy of the measurements.

In QuickStep scan, the XY scanner stops at each pixel point to record the data. It makes a fast jump between the pixel points.

High Throughput QuickStep Scan is ten times faster than conventional SCM scan with no compromise of signal sensitivity, spatial resolution, or data accuracy.

The PinPoint Conductive technology provides on-location electrical conductivity data at specific points on the sample. This means that users can now acquire contact current measurement at any specific location on a sample, at varying tip pressures, and at a much higher accuracy and precision than has been possible to-date.

Added benefits from this technology are reproducible data from repeated measurements, cost savings from longer lasting AFM probe tips, and sustained nano-resolution.

The comparison of conductive AFM images of ZnO nanorods show that the conventional contact conductive AFM may have a higher current measurement than tapping conductive AFM, but its resolution is compromised as the tip wears out in contact mode topography. The new PinPoint conductive AFM shows the best of both higher spatial resolution and optimized current measurement.

Conductive AFM is an important tool for device research and failure analysis. The conventional conductive AFM prevalent in the industry has to sacrifice spatial resolution in contact mode due to tip wear, or current level in tapping mode due to the short contact time. PinPoint Conductive AFM provides the best of both methods, with higher spatial resolution and optimized current measurement.

How do you see the AFM industry developing over the next few years as widespread adoption of the technique continues in a variety of fields?

In general, the AFM industry has transitioned from early adopters to an early majority market, and it is now moving into the mainstream scientific instrument market. This is attested by the recent acquisitions of AFM manufacturers by large conglomerate instruments manufacturers - Bruker acquired Veeco’s AFM unit in 2010, and Oxford Instruments recently acquired Asylum Research in 2013.

However, the rate of adoption of AFM instruments differs greatly from one market segment to another. AFM is now a must have instrument for nanoscale studies in the physical sciences. We feel this market will continue to grow steadily.

In the life science research market, AFM is still in the earlier part of the adoption curve. Most AFM studies in biology are limited to force-distance curves. This is due in part to the shortcomings of conventional AFMs – AFM tips can easily damage soft biological tissues when scanning in contact or tapping mode.

Some samples (e.g. micro fibrils) are so soft that they cannot be imaged by AFM, or any other instrument at all. Park’s scanning ion conductance microscopy (SICM) has made live cell imaging not only possible, but also practical. We believe that there will be a fast growth in this area as researchers in life science start learning and using this tool, and other innovations at the forefront of AFM.

As for the semiconductor market, there was a real struggle in that market to transition to widespread use of AFM technology. The reason for this was simply that earlier instruments could not perform the necessary functions (e.g. sidewall measurements).

Our 3D AFM technology provides these critical measurements – and we are beginning to be noticed in the market accordingly. We are sure that our success in this market will continue to grow as engineers learn of our instruments’ capabilities.

It’s an exciting time for our company and for the AFM industry.

About Dr Sang-il Park

Dr. Sang-il Park is the Chairman and CEO of Park Systems Corporation, a leading manufacturer of Atomic Force Microscope since 1997. Dr. Park founded Park Scientific Instruments, the first commercial manufacturer of Atomic Force Microscope, where he served as the Chairman and CEO for 9 years from 1988 to 1997.

Prior to founding Park Scientific, he worked with Prof. Cal Quate at Stanford University as a graduate student and research associate.

Dr. Park authored and co-authored numerous research papers, text books, and twenty some U.S. patents in the field of Atomic Force Microscope. He received his Ph.D. in applied physics from Stanford University and his B.S. in physics from Seoul National University.