Community Research and Development Information Service - CORDIS

FP7

AIM4NP Report Summary

Project ID: 309558
Funded under: FP7-NMP
Country: Netherlands

Final Report Summary - AIM4NP (Automated in-line Metrology for Nanoscale Production)

Executive Summary:
Nanometrology can add high value to production processes as it allows measuring and controlling the functionality of products, and reducing material and energy consumption. An essential prerequi-site to make these promises true is the integration of nanometrology into the production line.
The aim4np consortium took on that challenge and developed an instrument for tactile in-line me-trology. Focus was put on tactile measurements as they allow access to important parameters such as roughness, adhesion, elasticity, or hardness that are not directly accessible by e.g. optical means. The method must be fully functional in vibrating environments, tolerant to contamination, and work on large samples. The results must be traceable to international standards. Throughput or more precisely time-to-data are critical characteristics.
Our solution is based on robotic placement of an instrument, which allows quickly accessing the critical positions and limit time-consuming measurements to a handful of those points that had pre-viously been identified e.g. during ramp-up of the production process.
The instrument itself is designed as a generic metrology platform onto which different metrology heads can be mounted and which serves as reference plane for them. This platform can be at-tached e.g. to the afore-mentioned robot or a fixed frame depending on the requirements. The plat-form itself features a 6-degree-of-freedom [6DOF] displacement measuring system with which the relative position and motion of the sample can be recorded. A control-unite and a 6DOF actuator are used to actively stabilize the relative position between sample and metrology platform to a constant working point. This generates a stiff, virtual link between the reference plane of the metrology heads and the sample. The heads, which executes the nanomechanical measurements of texture and roughness, was a special atomic force microscope [AFM] with self-sensing cantilever-probes.
The 6-DOF instrument was successfully built and tested. AFM images with a resolution of ±3nm were achieved in an environment where production conditions were simulated during the measure-ment. Potential for further improvement was identified. The concept of robot-based measurements was demonstrated using a 1-degree-of-freedom version of the metrology platform.
A robust transfer standard and calibration methodology was conceived and successfully applied to calibrate the instrument and its measurements, making the results traceable to the standard of the meter. A process for calibrating the tip radius of an AFM-cantilever was studied and successfully tested. Such tips are delicate and if worn-out may influence the nanomechanical measurement re-sults. We therefore had researched and implemented a continuous tip-monitoring technique, that delivers a criterion for automatic tip exchange. A patent application was submitted for this method.
The application of nanomechanical characterisation was studied in more detail for two cases: plastic injection moulding and organic solar cells. Modelling and experimental validation of the injection process led a simulation tool for analysing nano-roughness in the production of macro-scale objects. For the case of solar cells, we concentrated on the transparent top-side electrode, which is fabri-cated from a conductive polymer filled with silver nanowires [Ag-NW]. We could establish the link between the Ag-NW concentration as measured in AFM and the electrical conductivity and optical transparency resp. Both are critical for the performance of the solar cell. The two models success-fully link the nanomechanical measurements to the functionality of the final product and could be used to optimize or control the production process in future.
Different elements of the instrument as the tracking sensors, AFM scanner, AFM canti¬levers, and the simulation software are independently valuable and the industrial partners of the project will start exploiting them in the near future. The academic partners identified open re¬search questions that could not be addressed within the current project, but for which new proposals had been submitted.
The project was presented at several national and international conferences and fairs. It so far led to 9 peer reviewed journal publications and 32 conference presentations. The more recent results will be disseminated after the end of the project.

Project Context and Objectives:
Summary Description of the Context and the Objectives of the Project

Production technology is on the verge of a profound change by entering the nanometre domain. In-line metrology at this scale is a pre-requisite for reliability, high yield and throughput. The aim4np project demonstrated for the first time tactile measurements of relevant nanomechanical properties of devices directly in an environment that suffers from vibrations typical for a production line.

Context

Increasing the resource efficiency by controlling nanomechanical properties during manu-facturing

Stiction and friction are at the origin of energy losses, reduced tool-lifetime, loss in precision and resource wasting in many manufactur¬ing processes. In order to control these effects in an efficient way, a reliable measure¬ment also of optically not accessible key nanomechanical parameters such as roughness, adhesion, or hardness, is needed. Moreover the produc¬tion process must become actively controlled at the nanometre scale, imposing the need to monitor these parame¬ters in con-tinuous fabrication processes, such as in Roll-to-Roll [R2R], Sheet-to-Sheet [S2S] production, to ensure high yields.
Nanoscience has delivered instruments such as the Atomic Force Microscope [AFM] and related scanning probe microscopes [SPM] to investigate exactly these characteristics with the required accuracy and spatial resolution. Their successful application in the laboratory environment has sig-nificantly increased our knowledge about the fundamen¬tal phenomena in nanoscale mechanics and their link to macroscopic observations. Comparable instru¬ments that are compatible with production environ¬ments, however, did not exist but are an absolute necessity.

Assessing relatively large surfaces in reasonable time – knowledge based, fast probing

A further, big and fundamental challenge in nanometre metrology is that real workpieces and ma-chine tools have macroscopic dimensions, while nanomechanical properties depend on features in the nanome¬tre domain. Good examples are flat-panel displays, which are manufactured on plates of 2 by 2 metre or larger while their functionality depends on thin films and circuits in nanometre do-main. Regardless of the measuring method, analysing the entire surface would require a huge dy-namic range of more than one in one million. A correspondingly large data-volume on the order of 100 TB/cm^2 over square meter sized surfaces would be generated. Even heavily parallel measure-ments and data-analysis for characterizing the entire macroscopic sur¬face at the nanoscale will never reach the neces¬sary time-to-data required for in-line controlling fabrication pro¬cesses, thus new concepts were needed here as well.
Micro-electronic industry has demonstrated that analysing the entire surface of a work-piece is not required for an efficient production. It achieves impressive yields of more than 99.9% even in highly com¬plex processes. In this exemplary approach, the process is carefully characterized during the development phase, while simultaneously critical positions are identified, which serve as proxy for the status of the whole work piece or tool. Later, during the actual production, only this handful of positions is probed. Comparably in other domains, e.g. for the roughness of moulds for plastic-injec-tion moulding, the deterioration of only a few, critical positions will be sufficient for detecting its life-time. Or in the case of solar cells fabrication, where transparent films are rendered conductive by suspending silver-nanowires, the content of nanowires has to be measured only at a few spots to ensure its electrical and optical quality before the processing is continued.
The objective of aim4np was to address these challenges, and to develop the required instruments and tracea¬ble measuring methods. In parallel, we wanted to establish the scientific background knowledge for a proof-of-concept study in a production line by linking primary nanomechanical properties to func¬tional requirements of products.

Objectives

Bringing nanomechanical measurements to the fabrication line

Most laboratory-type instruments accept only small samples, are vibration sensitive, and lack the flexibility and ease-of-use for applications in production environments. A paradigm shift was needed: It was the main objective of aim4np to bring the nanomechanical analysis of the tools and workpieces, directly into the fabrication environ¬ment i.e. to the vibra¬tion and noise of the work floor. Previous attempts to accomplish that by downscaling size and masses while increasing stiffness could not be implemented in full or did not lead to the required stability. For special applications in the IT industry enclosed ‘microenvironments’ were generated, which can be considered as a small, sealed-off laboratory brought to the fab. This approach is of limited broader use, in particular for continuous production systems.

Any undesired relative motion between the sample and the probe is affecting AFM and high precision optical measure¬ments. We eliminated these disturbances by mounting the instrument above the surface of the workpiece on a metrology platform [MP] with a near zero-stiffness suspension. Tracking and actively stabilizing the relative position between the platform and the workpiece by mechatronic means allowed locking the two together in a contactless loop, which can be considered as an artificially stiffened frame. This concept is schematically outlined in Figure 1.

A further objective of aim4np was to provide a metrology tool that can be flexibly positioned to workpieces and tools in the production line. This should be possible on large areas without dissecting or otherwise destructively manipulating the workpiece. The above-presented concept provides the freedom for integrating the instrument on a multi degree of freedom robot. The generic metrology platform can even approach and look to diffi¬cult-to-ac¬cess locations e.g. to the web in a R2R fabrication environment.

This fast placement of the instrument to the above-mentioned critical positions is the first step towards short time-to-data. In future, the instrument developed by aim4np can be positioned with high-speed robot, ideally comparable to pick-and-place machines in microelectronics. It is hence commensurate to very high volume processing. Even performing the measurement on a moving target (e.g. in a Rol-to-Rol process or on a conveyor belt) could be envisioned in future due to the inventive tracking and virtual stiffness solution.

Traceable measurements

Typical metrology instruments are operated in well-protected quality assurance laboratories, which guarantee stable working conditions. Once calibrated, such instruments provide traceable measurements, and their re-calibration is only needed after prolonged use. The situation in the production line might be completely different; hence, frequent recalibration is needed. It was a secondary goal of aim4np to develop a concept that enables calibration without only a short interruption of the measurement tasks and without the need to un-mount the instrument. For that purpose, aim4np has set the goal to develop a robust transfer standard, i.e. a standard that is calibrated against a primary standard and that is tolerant against contamination and (minor) damages of its surface during its use in the production line.

Autonomous operation

The outlined instrument must be functional without the need of a highly trained operator. There are a few key elements to achieve such autonomous operation: 1. Cantilever readout must not require optical alignment. For that purpose, aim4np had the goal to develop self-sensing cantilevers.

2. As the tip quality could affect the accuracy of the measurement, aim4np had the objective to develop a tip-monitoring technique that does not require interruption of the measurement process or use a special calibration sample.

Dependence of product functionality on nano mechanical measurements

It only makes sense to measure nanomechanical properties in a production line if their influence on the fabrication process or on the functionality of the product is understood. It was therefore an objective of aim4np to investigate this dependence for two special cases, which were selected for validation purposes. The goal was a description in the form of a mathematical model that could eventually be used for autonomous analyses of the data in future. The selected cases were plastic injection moulding and organic solar cells.

In plastic injection moulding we wanted to understand, how the surface roughness of the mould is transferred to the moulded piece and how this transfer can be influenced by production parameters such as e.g. pressure or temperature. The sought-after model should also be useful for deducing the roughness of the mould, incurred e.g. by wear, based on the roughness as observed on the moulded piece. This could then be used for evidence-based maintenance of the sometimes rather expensive moulds.

In the fabrication of organic solar cells there are many process steps that require careful nanomechanical measurements during the ramp-up phase. One critical element, that needs also to be monitored during production, is the electrical conductivity and optical transmittance of the top electrode. Optically transparent organic conductors suffer from a low conductivity. Adding silver nanowires [Ag-NW] to the precursor of the film helps increasing the conductivity but at the same time reduces the optical transparency. We wanted to understand whether surface texture measurements of the coated film can provide insight into electrical conductivity and optical transmission. The hypothesis was that we can deduce the concentration and dispersion of the Ag-NW from texture measurements. It is known how both parameters affect the functionality.

Project Results:
This chapter is structured into parts on
1) the main nanometrology instrument and its sub-components,
2) the calibration and tip monitoring,
3) the model forming for linking nanomechanical properties to functions of the product including the respective experiments, and
4) the validation experiments.

1. Nanometrology Instrument

Environmental Conditions and Requirements

The vibration amplitude, acceleration and spectrum, the temperature and relative humidity were measured in a relevant fabrication environment at a partner’s site. From these measurements the following performance requirements for the aim4np instrument were derived (c.f. also Figure 2)

Vibrations

The vibrational spectrum shows frequencies up to about 500Hz with relevant disturbances up to 200Hz (c.f. Figure 3)
Maximum peak-to-peak displacement: 19 µm (robot arm)
Velocity spectrum, highest peaks: 245µm/s @ 50 Hz (robot arm)
60µm/s @ 97 Hz (floor)
150µm/s @ 120 Hz (preparation table)
Acceleration spectrum, highest peaks: 80mm/s^2 @ 50 Hz (robot arm)
120mm/s^2 @ 120 Hz (preparation table)
Maximum peak acceleration 3.1m/s^2 (preparation table)

Actuator

The prime task of the actuator is block transmission of vibrations by counter-acting them. Hence it needs to have:
Actuation force: 1.5N (rms) and 13N (peak)
Actuation range: > 25µm

Sensor

The sensor-system measures the relative motions between the reference frame and the sample. For that it needs a compact design and provide in- and out-of-plane measurements with:
Bandwidth: >5kHz
Measurement range: few µm
Sensor noise: single nanometre

Traceable measurements of the surface texture (roughness and topography) in the sub 100nm range are not common. None of the standardization laboratories worldwide have reported such Sq values up to now. The following targeted values are based on optical measurements (for injection moulds) on a sample that showed the required end performance, and on estimations based on thin-film growth (solar cell). The tolerance values (± X nm) are to be interpreted as hard (100%) intervals.

For injection moulds / moulded pieces:
Areal Surface Texture according to ISO 25178-2:2012 Sq = 20 nm ± 5nm

For thin film organic solar cells:
Areal Surface Texture according to ISO 25178-2:2012 Sq = 10 nm ± 3nm

Architecture and systems design

We succeeded in compensating or rejecting the above mentioned disturbances with a system that measures and actively stabilizes the relative position to the workpiece in 6 degree of freedom [6DOF].

Our engineering research led to the following modular concept: A metrology platform [MP] is suspended from a mounting point by means of a (near) zero stiffness suspension that can be attached to a robot or fixed frame. The suspension is combined with passive gravity compensation and a 6DOF tracking actuator that can move the metrology platform. The MP is equipped with tracking sensors to measure the relative position in 6DOF. These sensor signals (S1, ..., S6) are used to calculate the actuation signals in a multi-input-multi-output [MIMO] feedback loop in order to exert the forces (F1, ..., F6) that stabilize the position of the MP (c.f. Figure 4). The electromagnetic tracking actuator is based on the Lorentz principle and combines all DOF in one compact unit. It also establishes the gravity compensation by means of permanent magnets. The three vertical tracking sensors are optical sensors based on the Foucault (or focus) detector principle. The lateral tracking sensors are also optical sensors, however, based on the heterodyne detection of light reflected from the moving surface. Alignment of the different sensors was critical.

The MH was designed as monolithic aluminium frame onto which the different tracking elements can be bolted. A CAD drawing of the system is shown in Figure 5a) next to a photo from the real system in operation at the test-stand.

The proper measurement or metrology heads [MH] are also attached to the MP, which serves as reference for their measurements. There were two types of MHs built: an atomic force microscope [AFM] scanner and a white light interferometer. The approach of the AFM sensor to the sample is provided by a separate stage, which is part of the AFM-scanner. Mass balancing and the dynamics of the populated MP were critical and required a careful design.

Tracking Actuator

The newly developed, compact tracking actuator comprises two main elements, the stator which is attached to the robot or fixed frame and the mover to which the MP is attached.

The implemented gravity compensation is based on a design proposed by S. Hol [S.A.J. Hol, Design and optimization of a magnetic gravity compensator; PhD thesis, Eindhoven University of Technology, 2004] , which relies on two concentric, magnetic tubes, one (the mover) with radial the other (the stator) with axial magnetization. This provides zero stiffness at the working point and a stiffness of <10N/m in a 4mm working range of the suspension.

Displacement of the mover is achieved by dynamically superimposing the permanent magnetic field with electrically generated ones. Six specially wound coils are mounted on the stator for that purpose (c.f. Figure 6). The generated forces (motor constant ~0.9 N/A) allowed accelerating the 4kg MP [4MP] within the required displacement and dynamic range. Especially at high control-bandwidth the power consumption is significant and thermal design became critical. The design solution we found was sufficiently cooled by thermal conductivity through the structural elements.

Tracking Sensors

The vertical tracking senor measures displacement in z-direction. Sensing is based on the astigmatic behaviour of the imaging optics due to changes of the stand-off distance of the sensor. The sensor consists of the hologram unit with implemented Foucault detector, which allows detecting the distance-change to the sensor around the focal plane of the lens. Three of these vertical sensors were combined into one unit together with the required electronics (c.f. Figure 7a). This allowed also measuring the tilt of the sample plane relative to the reference plane of the MP.

A comparative measurement (c.f. Figure 8a) with a laser-interferometer demonstrated that the required nanometre resolution has been achieved. The noise is below 2 nm peak-to-peak measured at full bandwidth. The resolution was found to be 2.6µm. The bandwidth of the vertical tracking sensor is above 8.5 kHz, and the measuring range ±1.5µm.
The lateral or in-plane tracking sensor is based on the heterodyne detection of the moving sample surface in the diffusively reflected light of two laser beams of different wavelength, shone under an angle to a common point on the surface (c.f. Figure 7b). Two PLL-coupled He-Ne lasers have been used as light sources. The frequency difference for the lasers can be set in the range between 2 MHz and 7 MHz, which allowed simple solution for the signal demodulation because of the relatively low demodulation frequency.
The noise of the lateral tracking sensor is below 15 nm peak-to-peak at full bandwidth (c.f. Figure 8b). The observed global drift in Figure 8b is caused by the thermal drift of the setup. The resolution was found to be 5nm. The bandwidth of the vertical tracking sensor is above 10 kHz, and the measuring range ±10µm.

Performance of the Metrology Platform

Successful operation of the MH was possible after careful system identification and tuning of the feedback loop. Figure 9 shows the residual tracking error in 6DOF when the MH is kept at the working point of the AFM. These tracking errors are small enough to perform AFM imaging as will be demonstrated later in this report. The tracking error of 17nm (peak to valley) in vertical direction is about an order of magnitude smaller than the 5% of the AFM actuation range (37µm) that was specified in the Description of Work. Table 1 shows the tracking errors that were achieved.

Optical Metrology Head

The highest stability white light interferometer [WLI] for industrial application can be achieved with the Mirau principle, because the reference mirror is included into the interferometer’s objective in this arrangement and there are no moving parts needed. The WLI selected for aim4np is based on a commercial product of partner SIOS. The dynamical behaviour of the body was analysed and the lowest resonance was found at 433Hz and was caused by the camera system of the device.
The optical metrology head has a field of view of 370µm x 470µm, and is able to measure the topography (height) of the surface with uncertainty below 4 nm and resolution below 1 nm. The lateral resolution is given by the diffraction limit.

Scanning Probe Microscope - AFM Metrology Head

The AFM metrology head is used for measuring the texture of the workpieces. In contrast to a normal situation, the use within aim4np purposefully exposes the AFM scanner to dynamic inertial forces during sample tracking motions of the Measurement-Platform. This demands for a miniature, low mass, fast scanner with position sensors and closed-loop operation.

Different design possibilities have been investigated and evaluated. Nanosurf decided to develop and build a knuckle-joint type flexure structure for the xy-scanner and a Chevron type, counterbalanced flexure structure for the z-scanner, that are small and lightweight (35x23x40 mm, 30 g) in order to tackle the demanding environmental requirements.

The new scanner (see Figure 10) has about 20 times less volume and weight than current Nanosurf AFM scanners, it is fully piezo-actuated, and is approximately 50 times stiffer and accordingly allows higher scan speeds: the Chevron-type z-scanner has its first z-resonance at 8 kHz at a remarkably high range of 30 µm.

Nanosurf has developed closed-loop control code (FPGA) that can deal with the lowest mechanical resonances of the respective scanner axes to increase closed-loop bandwidth and reject excitations by the tracking motion up to higher frequencies.

Miniature optical position sensors have been used and improved for the operation in the aim4np AFM. Especially a miniature optical 2D sensor with low noise (<0.5nm rms) has been developed for the 100 µm xy-scanner for closed-loop operation to deal with the coerced vibrations generated by the tracking motion. The developed aim4np-AFM fulfils the following specifications:
• Lateral scan range of 100 µm.
• Vertical scan range of 30 µm.
• Height resolution below 0.5 nm (z-noise floor <0.2 nm rms).
• Lateral resolution below 5 nm (noise floor, open loop: < 0.5 nm rms, closed loop: < 2 nm rms).
• Image acquisition line rate up to 20 Hz.
• Closed-loop operation in all axes.
• Use of easy-to-use self-sensing cantilevers

The AFM is operated with easy-to-use piezo-resistive self-sensing cantilevers, which avoids the tedious laser-alignment process needed with laser-deflection sensing. These cantilevers were developed by partners NanoWorld Services and Nanotools.
Figure 11 shows two examples of images acquired with the AFM on a standard AFM support in closed-loop operation for slow (600ms / line) and fast (50ms / line) scanning. More images are shown in Chapter 4 for the AFM operated on the Metrology Platform.

2. Calibration and Tip Monitoring

Introduction

Reliability of measurement results is crucial to unambiguous interpretation and quality assessment of both production lines and the products that are produced. Given the metrology platform itself and the two types of instrumentation on the metrology platform, i.e. a white light interferometer [WLI] and atomic force microscope [AFM], the development of traceability for AFM became the main focus since existing methods to provide traceability for the platform and for the WLI were considered to be sufficient for the current project. The main challenge to provide traceability for AFM within aim4np is given by the industrial setting of the metrology platform. Existing methods were considered not to provide sufficient robustness, flexibility and accuracy for such a setting.

The areal surface roughness parameter Sq was specifically selected as indicator for the functionality of the surfaces of the aim4np products, i.e. moulds and solar cells. Methods for reliable measurements of Sq were developed by VSL.

Overall traceability strategy

Within the first month of the project, an analysis was made to define the conditions for providing traceability to the system based on the system concept. Potential sources of measurement uncertainty were defined and a traceability strategy based on calibration standards was selected Figure 12.

The areal surface roughness Sq was considered the most suitable parameter within the aim4np project for assessing product quality and functionality. Given the lack of respective commercially available standards, VSL has developed a concept to realize a practical standard, Figure 13.

Instead of a physical implementation of such a standard, the concept was tested successfully using a programmable surface profile generated by means of a piezoelectric element. This concept was later generalized into virtual standards for calibrating the measurement space of other AFMs.

Virtual standards

In order to provide accurate calibration capabilities to in instrument in an industrial setting several conditions have to be considered. For once, physical standards do not provide an acceptable solution because of the delicate nature of such standards; the nanostructured surfaces of these standards are very susceptible to contamination and damage in a harsh environment. Additionally, several physical standards are required to calibrate the range of values that will be typically used during normal operation. The effort to calibrate the instrument therefore scales with the number of standards that are required. For an industrial setting this is highly undesirable. Finally, the metrology platform in the aim4np project is designed to perform measurements on the nanoscale with sub-nanometre accuracy requirements. The availability of physical standards to that level is limited or non-existent depending on the required parameter.

The concept of virtual standards has been implemented using a special type of piezo material that exhibits low hysteresis and low non-linearity. Two types of standards were developed; one for the calibration of the height axis of the AFM and one for the calibration of both lateral axes, Figure 14. These standards provide a programmable range, are robust against contamination and damage and offer sub-nanometre accuracy.

Calibration of the virtual standards

Calibration of the piezo translation down to the picometer level involving displacements from 2 pm to 20 nm was achieved by using a recently devel-oped, dedicated calibration facil¬ity at VSL . The calibration result showed (c.f. Figure 15) that the behav¬iour of both the height standard and the lateral standard was ex¬tremely linear and accurate and there¬fore well suited for their intended purpose. The estimated measure¬ment uncertainty for the sensitivity coefficients of both standards is 0.5 %.

Application of the virtual standards

In order to apply the virtual standard for the calibration of the AFM, measurement and analysis methods were developed to provide the best possible measurement uncertainty while maintaining easy to use functionality that could be applied in a production environment.

The VSL Dimension AFM has been used as a test bed to establish the full potential of the virtual standards. The standards have been used to calibrate the height and lateral axes, establish and correct scanner non-linearity and determine the squareness errors between the axes and finally, validate the measurement performance for the Sq areal surface roughness parameter, (Figure 16).

The Icon AFM of CSIC was calibrated in the height and in one of the lateral directions, as part of the first field test with the virtual standards. During the validation experiments in Vienna, one of the Nanosurf AFMs used in the aim4np instrument was calibrated in the height direction, (Figure 17).
The results of these calibration activities have shown that the actual sensitivity can differ up to about 20 % compared to the default values supplied by the manufacturer. The standards therefore not only provide traceability but also offer the opportunity to increase the accuracy significantly when the calibrated sensitivity values are used.
The standards are now routinely used within VSL to provide formal traceability for our AFM measurement services.

Uncertainty analysis

Based on results from the calibration of the virtual standard, the VSL AFM calibration and the performance results from the aim4np metrology platform, several uncertainty budgets were calculated to estimate the measurement uncertainty for the AFM measurement space and the uncertainty of the probe shape parameters, i.e. radius and cone angle.

The measurement uncertainty of the interferometer system used to calibrate the vertical and lateral tracking sensors in air is about 0.13 ppm and therefore negligible compared to the noise of the sensors. The resulting 3D position uncertainty of the platform due to the performance of the sensors is estimated to be 11 nm. Details are given in Table 2 and 3 respectively.

Probe shape calibration strategy and implementation

Several methods exist to reconstruct the shape of the AFM probe in order to correct the raw measurement data for probe-shape effects. Blind reconstruction techniques and methods based on measurements of sharp features are used often but were not considered appropriate since these techniques and methods lack a sound traceability basis. We have decided to use well-defined round nanoparticles as features that could be used as a suitable means to measure the probe shape in a traceable manner. The roundness of the particles ensures that the height is equal to the width of the particles.

Suitable nanoparticle samples were found not to be available so the preparation methods had to be developed in house at VSL. As a result, we now have recipes to produce samples with a wide range of particle diameters (Figure 18) and with different nanoparticle materials like polystyrene, silica and gold.

AFM measurements on deposited round nanoparticles will be distorted due to broadening of the profiles by the finite probe shape but in a well-defined way: Since the height of the particles can be relatively easily and accurately measured in a traceable way, the real particle diameter can be established. Measured deviations from the actual diameter in lateral direction are due to the probe shape and allow accurate reconstruction of the probe shape, equivalent to a traceable calibration. Once the probe shape is known from calibration on the nanoparticle samples, the raw measurement data on other samples can be corrected in order to increase the accuracy of the measured data, (Figure 19).

Tip monitoring

The AFM cantilever tip can deteriorate during measurements, which affects the result and could mislead the interpretation. We, therefore, developed a method to continuously monitor the physical condition of cantilever tip while AFM imaging in dynamic mode (amplitude modulation). The technique is based on the dynamics of the cantilever and the generation of higher harmonic vibrations due to the nonlinear interactions between the tip and the sample surface. This nonlinearity depends among others on the tip radius. Hence the increasing amplitude of higher harmonics can be used as indicator for an increasing tip radius.
We have applied this method to microfabricated silicon cantilevers in the 3-50 N/m force constant range and in the 50-400 kHz frequency range with ultra-sharp tips (initial tip radius below 10 nm) focusing in the resonance of the 2nd flexural mode and the 6th harmonic. We have observed both experimentally and by means of simulations using the Virtual Environment for Dynamic AFM (VEDA) open code that the amplitude of the 6th harmonic, averaged over the amplitude-images and acquired simultaneously with the topography and phase images, monotonically increases for increasing tip radius as shown in Figure 20.
An European patent was filed and received good reviews from the examiners.

3. Linking Nanomechanical Properties to Functions of the Product - Model Formation

Correlating roughness and texture with the performance of plastic injection moulding

The applicability range of commercial simulation codes devoted to plastic injection, which is typi¬cally above the millimetre range, has been extended to the nanometre domain. Such improvement permits for the first time the prediction of the replication of the surface topography of the mould surfaces into the moulded plastic pieces. The implementation of struc¬tured (both random and patterned) surfaces has been achieved by using submodelling, which con¬sists in simulating with a coarse mesh the part at mm scale and use outputs from such simulations as inputs for simulations of surface effects with a finer mesh at nm scale. This approximation has been proven to be successful in describing the actual morphology of diamond-like coated films used as protection of stainless steel surfaces of moulds as well as in silicon-based gratings with tar¬geted geometries.

A diagram for the submodelling is shown in Figure 21. In the first step it is assumed that the cavity has no influence on the propagation of the fluid because of the short time the propagating polymer melt needs to cross the cavity (~ 150 µs). From this macro simulation, the pressure profile and the melt front temperature of the polymer can be extracted. In the second step the pressure pushes the polymer into the cavity. This step ends when the polymer solidifies due to the heat transfer to the walls. The nanosimula¬tion is stopped when the polymer reaches its glass transition temperature in the entire cavity. ANSYS FLUENT and MATLAB were used for the simula¬tions.

Figure 22 shows the comparison of the predicted and experimental Sq values corresponding to polycarbonate injected pieces. From the figure it can be observed that the experimental values exhibit lower roughness values, in line with the prediction of the simulations.

Correlating morphology and performance organic solar cells

The top electrode of an organic solar cell must be optically transparent and have a good electrical conductivity. The later is achieved by blending silver nanowires [Ag-NW] in the conductive polymer that forms the matrix of this thin-film electrode. If the Ag-NW content is too high, the optical transparency is reduced and the performance of the organic solar cell worsens. We found that the coverage rate with Ag-NW can clearly be measured by means of AFM. The correlation of coverage with respectively the transparency and conductivity can be gauged by independent measurements (c.f. Figure 23) e.g. during ramp-up of the fabrication process. Hence, the AFM measurement can be used to assess the quality of the transparent conductive electrode in the production line.

In our experiments we have used poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) as polymer. Both the PEDOT:PSS solution (Clevios PH1000) and the Ag-NW solution (Sigma Aldrich 0,5% Ag nanowires in isopropyl) were used as purchased. Blend solutions with different ratios of PEDOT:PSS-solution/Ag-NWs-solution (and 10% of Dimethylsulfoxide (DMSO)) were spin-coated (3000rpm, 60s) onto pre-cleaned glass substrates and subsequently dried on a hotplate (150°C, 15min). UV/VIS transmission spectra were taken using a Perkin Elmer lamda900 spectrometer. For the conductivity measurements (4-point-probe) four silver electrodes were thermally evaporated on top of the spin cast layer by using a homemade evaporation chamber. The dimensions of the electrodes were determined by means of optical microscopy. 4-point-probe conductivity measurements were carried out with a Keithley Source Measure Unit. The determination of the nanowire coverage rate was executed with a Veeco Dimension V AFM controlled by a Nanoscope V controller. The same samples were later again analysed with the aim4np instrument (see Chapter 4).

4. Validation Experiments

Configurations of the instrument, experimental set-up

Various instruments and component were built in the course of the project, during which we drove the TRL form about 2 to somewhere between 5 and 6. All AFM images reported in this chapter were recorded with the instruments developed within this project otherwise noted.

There were two different experimental implementations used to conduct the AFM experiments: First a configuration where a 6 degree-of-freedom [6DOF] MP with AFM I was mounted to a rigid lab-frame, second a configuration where a 1 degree-of-freedom [1DOF] MP with AFM II was mounted on a portal robot. Most of the experiments were performed in the first configuration. The achieved results are compared to measurements made with a high-end laboratory AFM ) and in one case (plastic injection-moulded grating) with a confocal optical microscope. The instruments we used to validate and test the aim4np concepts in a simulated operational environment were:
AFM I & II: 2 AFM scanners of almost identical performance, built for redundancy
Robot: Four-axes Philips CAT robot of which however only two (x1 and y1) were used. A motorized vertical stage
(X- VSR20A, Zaber Technologies Inc., Vancouver, British Columbia, Canada), was installed as z-axis for
approaching the 1MP to the sample.
Shaker: A 1D - shaker (TV 51 110 of TIRA Schwingtechnik, Germany) used to excite vibrations comparable
to the one typically encountered in a production environment.

Calibration

The calibration of the height axis of the Nanosurf AFM was performed with the virtual standard for different settings of the step height. The Nanosurf system acquires images line-by-line without stable timing between the lines while the timing of the virtual standard signals is more accurate. As a result, the positions of the steps as they appear in the acquired images fluctuated. Nevertheless, the developed analysis methods could be applied to parts of the images. The measured step heights were calculated from the histograms of (parts of) the images. The applied steps in the Z direction result in two distinct height levels that show up as peaks in the histogram of this data. For the 10 nm step height, a clear separation of the histogram peaks representing the lower and upper level was observed. For the 1, 2 and 5 nm step heights only a single histogram peak for the raw data was visible indicating that the noise level was comparable to the step height. In order to extract the step height from these data sets, a median filter was applied to reduce the effect of noise while preserving the over¬all levels. The calculated calibration coefficient for all step heights is consistently about 0.8, (see Table 4). The actual Z displacement is therefore lower than the AFM indicates. An indication of the noise level can be extracted from the histogram in Figure 25. The half width at half maximum is about 4.3 nm, correction with the proper calibration yields 3.4 nm.

Effectiveness of the aim4np concept

Two different controller settings were used in order to demonstrate how effective the new technology rejects disturbances. The controller cannot be fully switched off, because otherwise the instrument would drift out of its labile balance given by the magnetic suspension and gravity compensation. A relatively low controller bandwidth is sufficient to keep the MP in a stable position. At this setting, however, disturbances cannot be compensated. The difference between the noise in AFM images recorded with a low and a high MP-controller bandwidth [BW] indicates the effectiveness of our technique. This is demonstrated in Figure 26, where a thin-film sample of silver nanowires blended in a conductive polymer matrix (c.f. page 22) was imaged. While scanning from bottom to top, BW was sent to 200 Hz and the shaker was off. After a few scan lines (~ 4µm in y-direction), the shaker was switched on to produce production-line equivalent floor vibrations. This caused well observable disturbances in the AFM image. After a few scan lines the shaker was switched off and the BW was increased to 400 Hz. Yet a few scan lines later, the shaker was switched on again. The effective noise-suppression can be seen by comparing the date with the final few scan lines, where the shaker was switch off again.

Plastic injection moulded samples

The experiments were performed on two different samples, one (Figure 27) with period grating (sample id: d5) of pitch 0.9µm, line separation 0.3µm and line width of 0.5µm and one (Figure 28) with an aperiodic grating (sample id: e5). For comparison, the same samples were also imaged using a confocal optical microscope (c.f. Figure 29).

Validation of operation on a robotic platform

The 1 degree-of-freedom [1DOF] MP with the assembled AFM II was mounted on the portal robot (CAT, Philips) in order to show the functionality of the concept on large samples, where the metrology head must be positioned at specific locations. We used a sample that was specially prepared by lithographic means (c.f. Figure 30), and that featured different, labelled alignment marks. The mask was accidentally mirrored during fabrication, which explains the mirrored characters in the AFM image (Figure 31a) and b)). The AFM was first use to register to one mark or label, then the robot moved to the location of the next mark, where the AFM took again an image.

Conclusion

We could successfully perform AFM measurements in an emulated production environment. The instrument used during these experiments was an advanced prototype well beyond a breadboard model, which however will still need to be further improved in terms of user-friendliness and handling as well as development of software support for real industrial applications. Nonetheless, the fundamental concept, key components and their effectiveness were validated.

The virtual transfer standard proved its effectiveness and in-sensitivity against contamination. It is ready for use also outside of the project.

The extraction of relevant production parameters from the performed nanomechanical measurements was shown for particular applications. There are now generic algorithms for this purpose and other applications will require their own, new modelling activities to support the interpretation of data.

Potential Impact:
Context

During the last decencies of the previous century, the policy of many European countries was to develop into a service society or to at least substantially increase services over manufacturing. In combination with the pull of low wages in developing countries, this lead to a proper de-industrialization in many segments. An essential exception was the high-tech part of the machine industry, the high-tech systems industry, which his highly knowledge driven and which requires highly educated and skilled workers. European countries are still among the world-leaders in this sometimes also as ‘Smart Industry’ referred segment. This domain is therefore a natural choice for starting the reindustrialization of Europe.

Smart Industry is characterized by highly knowledge driven solutions. Precise measurements, and accurate analysis and interpretations thereof are essential for reaching the aspired energy and material efficient processing. Basic science, e.g. nanoscience, has investigated many of the relevant phenomena and developed scientific instruments and sensing principles to measure them in a laboratory environment. Advancing these instruments into in-line equipment promotes the reindustrialization of Europe two-fold: First, it opens a new family of high-tech systems equipped with highly innovative metrology capabilities, and second, it allows effective manufacturing also in high-wage countries due to high added value of the corresponding processes.
The impact of aim4np has to be seen in this larger context: it provides a key competence of performing high-precision, vibration sensitive measurements at the nanometre scale level even in an environment that is by nature prone to vibrations. The application is not limited to tactile measurements, the solution that aim4np investigated, and will ultimately lead to new production processes, new production tools and lines and by this help reducing energy and material consumption and increasing the number of high-value jobs in Europe.

Profiling of current and near-future impact

The currently achieved impact is of course less wide and not very pronounced yet. It is focused on the market of the participating SME partners:
• Surface Profiler Market,
• AFM Instrument Market,
• Plastic Injection Moulds Market, and
• Solar Cells Market.
The application field is in high-precision and high-value manufacturing processes, where pure optical methods cannot provide the required information as they are not accessible to light, like hardness or adhesion of a surface, or due to wavelength limitations.

Strengths and Weaknesses

The so-called metrology platform, which aim4np developed, is a generic element onto which different metrology heads can be mounted. Each of these heads will be virtually linked to the workpiece, emulating a rigid, stiff reference plane. This modularity is a clear strength of the aim4np concept.

The innovative calibration process that we developed has demonstrated that measurements executed by the aim4np-instrument are traceable to international standards. This is an additional strength as it provides absolute values, which helps distributing production over different machines or even sites with guaranteed conditions and properties.

A weakness is the estimated cost for the instrument, which is expected to be too high (>100 k€) for applications in low-cost high-volume manufacturing processes. Similarly, aim4np cannot compete in applications where pure optical solutions exist, as they are considered to be more mature, faster and less complicated, i.e. where no paradigm shift is needed.

Further, the new instrument cannot compete with classic solutions, where sample and instrument can be placed on an anti-vibration table, as they again do not ask for a paradigm change and classic quality control approaches are sufficient.
TRL of the current solution is about 5 - 6 . A proper assessment is only possible after having identified the respective market. At this moment, the consortium has not yet found an integration partner who would be willing to further develop the instrument. The difficulty in finding such a partner is two-folded, first, he would still need to invest time and effort for reaching a product, second, there is no ‘killer application’. This is often the case for solutions that involve a paradigm-shift.

Opportunities

In the course of the project, the production of flat panel displays and touch screens were identified as manufacturing processes with high potential impact. Moreover, large optical elements (gratings, lenses, lens-arrays) and hybrid MEMS devices (sensors) are considered. One of the consortium partners (NS) received encouraging expression of interest from a glass manufacturer for measuring the nanoroughness on large (~ 3m x 3m) glass-panels. Even though flat panel displays form a large market, there are only few manufacturers who would benefit from the offered solution.

The AFM market was blocked for piezo-resistive self-sensing cantilevers by a patent of Stanford University for years. This patent has expired and the new high-frequency, self-sensing cantilevers developed by partner NWS are expected to find large interest beyond the application in aim4np. Operating them requires either some adaptation of existing AFM-scanners or using the new AFM-scanner developed by partner NS. Hence, this scanner is also expected to find large interest especially for automated applications and OEM solutions. A nice side-product is the method of tip-radius monitoring, which has been jointly developed and patented by partner TUD and CSIC. Exploitation through licensing agreements with different AFM-manufacturers is considered.

The calibration of AFM was not considered in the initial impact analyses because the call only asked to develop concepts for making the measurements traceable without actually demonstrating these concepts. The aim4np concept, however, turned out to be inherently very robust against contamination, a typical problem in production lines. It was so attractive that we advanced it further: A standard-operation-process has been developed which can now be used to calibrate AFM-instrument and other topography-profiling instruments. This is a niche market where partner VSL is active.

Plastic Injection Moulding maybe less interested in the instrument, as forecasted cost of ownership is too high for this low-price segment technology (see weaknesses above). However, the software module for modelling high-fidelity reproduction of surface texture during injection moulding processes, developed by partner IQS, appears to become an attractive side-product of the project.

The situation for the aim4np-instrument in Solar Cells manufacturing is not yet conclusive. A key issue will again be the cost-of-ownership. Potential applications are e.g. in the process-step where transparent electrodes are fabricated and optimized by blending Ag-nanowires to a transparent but poorly conductive polymer to increase its conductivity.

Expected impact for project partners

The perspectives for all participating SME partners to exploit results of this SME-targeted project are very good. Their first products and services will be based on elements and partial technical innovations needed to implement the full metrology instrument. An overview is given in Table 5.

The originally foreseen integration partner Anton Paar had decided to leave the consortium early in the project due to strategic re-orientation. For exploiting the complete instrument and the newly developed methodologies we are, hence, currently analysing two paths: first the creation of a start-up company into which the key results could be spun off, or second, a partnership with an OEM or one of the big prime-contractors for complex production lines like ABB, Bosch or Siemens. The OEM solution bears higher economic risks. The first solution would provide more time for introducing and demonstrating the readiness level of this paradigm-shifting technology. In order to increase the TRL, TUW, an Austrian SME and TUD have, jointly submitted a follow-up proposal to an Austrian funding agency (FFG) that supports pre-competitive developments.

Impact on workforce

The consortium partners report the following increase in temporal or structural workforce which are entirely or to a very large extend caused by the project:
Nanosurf: +1fte, structural
VSL: +1 fte, structural
TU Delft: +2 PhD students (temporal), 2 MSc students (temporal)
IQS: +1 PhD student (temporal), 1 MSc students (temporal)
TU Wien: +1 full PhD student (temporal), 2 part PhD student (temporal), 5 MSc students (temporal), 3 semester Thesis,
3 Postdocs (temporal).
CSIC: +1 PhD student (temporal),
Flubetech: +1 fte, structural
SIOS: expected 1 fte (structural) within one year after end of the project.

Conclusion

The call stated as expected impact:
"The results will lead to radical innovation in the design of products and production processes and to improving the performance of nanostructured coatings, rationalising industrial material selection, and boosting the competitiveness of the product manufacturers. In the long term, this will lead to enhanced operational performance of products, with impact on both competitiveness and sustainability. Measurement standards are a prerequisite to bring on-line the tooling technologies required to produce these products. The end users will benefit through lower energy bills and increased robustness of manufacturing systems. Also in the process industry, manufacturers will benefit from lower transportation costs. The results will also create new product opportunities for European instrumentation industry."

We conclude that most of these expectations have been accomplished or will be accomplished within the 1-3 years after the end of aim4np.

The new methodology to virtually lock the work-piece to the metrology instrument introduced by aim4np and implemented in a demonstrator represents a paradigm shift, hence a radical innovative design. Its active suppression of environmental nuisances enables a new generation of standard operation processes where crucial nanomechanical characteristics such as nanoroughness or adhesion are constantly monitored and fed back to control manufacturing of high performance products.

The developed models for linking the measured nanomechanical properties to the final functionality of the product enable radical new design processes and help controlling the fabrication process. In connection with the inline metrology, they will undoubtedly lead to quick, less material and energy consuming development processes and new standard operation processes.

aim4np has established a new, robust transfer-standard for in-line metrology. Together with the inventive technique to monitor the status of the crucial tip-radius of the AFM-probe, this standard ensures traceability for optical en mechanical roughness measurements.

The results produced en route to the final project goal constitute already exploitable side-products on their own and contribute to the competitiveness of the participating SMEs (see also Table 5). All together, the results of aim4np help to increase the competiveness of European manufacturers.

List of Websites:
www.aim4np.eu

There are links for the websites of each partner at the ‘Consortium Description’ page. Contact details for the different partners are:

U. Staufer Precision and Microsystem Engineering
TU Delft
Mekelweg 2
2628 CD Delft
The Netherlands

G. Schitter Instituts für Automatisierungs- und Regelungstechnik
TU Wien
Gusshausstrasse 27-29,
1040 Wien, Austria

J. Fraxedas CSIC CIN2
Campus de la Universitat Autònoma de Barcelona
08193 Berllaterra, Spain

A. Blümel Joanneum Forschungsgesellschaft mbH
Franz-Pichler-Straße 32
8160 Weiz, Austria

M. van Veghel VSL
Thijsseweg 11
2629 JA Delft, The Netherlands

A. Lieb Nanosurf AG
Gräubernstrasse 12-14
4410 Liestal, Switzerland

W. Schott SIOS Messtechnik GmbH
Am Vogelherd 46
98693 Ilmenau, Germany

T. Sulzbach Naonworld Services GmgH
Schottkystraße 10
91058 Erlangen, Germany

C. Kusch Nanotools GmbH
Reichnbachstr. 33
80469 Munich Germany

C. Colominas Flubetech, S.L.
Carrer Montsià, 23
Pol.Ind.Can Carner
08211 Castellar del Vallès (Barcelona) Spain

A. García-Granada IQS-URL
Vía Augusta, 390
08017 Barcelona Spain

Related information

Reported by

TECHNISCHE UNIVERSITEIT DELFT
Netherlands
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top