AFM Exclusive

10 Jul 2016

In Electrolyte Solution Topography Imaging of Various Organic Samples using Park NX10 Scanning Ion Conductance Microscopy

In order to identify and understand the root cause of a problem then develop an appropriate countermeasure, it is sometimes necessary to characterize a sample in situ. Several cutting edge applications follow this necessity such as: 1) the study of failure mechanisms in battery electrodes, 2) the study of membrane transport of ions between cells, and 3) the study of static and dynamic biomechanics at the cellular level. Common to all these applications is the need to perform in situ measurements of a sample experiencing a phenomenon while in an electrolyte solution. Techniques such as Atomic Force Microscopy (AFM) have been used for inliquid imaging of samples before, but AFM cannot be considered an ideal solution.

4 Feb 2016

Automatic Defect Review AFM with Enhanced Vision

By Dr. Ardavan Zandiatashbar, Ph.D., Research Scientist Park Systems Inc., Santa Clara, California.   ESSENTIAL TOOL FOR HARD DISK MEDIA FAILURE ANALYSIS Failure analysis engineers in data storage rely on atomic force microscope (AFM) as one of the key techniques for obtaining threedimensional information about the topological defects on hard disk media samples. This information is critical for proper identification of defects and eliminating defect sources. Although optical microscopy can provide a quick image of sample surface, the technique has limited resolution due to light wavelength. On the other hand, electron microscopy can provide higher resolution but it is essentially a destructive technique and provides only a twodimensional image. AFM provides the highest vertical and lateral resolution and provides a three-dimensional topography of the sample surface [1]. The conventional AFMs were known to have limited throughput and tip life since imaging was performed in destructive mode. To address these limitations of AFM, Park Automatic Defect Review (ADR) AFM has been developed and introduced in mid 2000s for a non-destructive high throughput 3D imaging of defects. This solution was widely accepted and welcomed in failure analysis (FA) labs globally. In this article we introduce the latest generation of ADR-AFM. The new ADR-AFM utilizes enhanced vision technique to facilitate locating defect of interest (DOI). This solution carries the commercial name of Park NX-HDM. The main objective of the FA lab for hard disk media is studying the defects and investigating the possible defect sources to improve yield management. For this purpose, defect study consists of two main steps: defect inspection, and defect review. In defect inspection, an inspection tool searches the media surface using optical, electron beam, or magnetic probe to find the location of the defects. The output is a map of defects which indicates their distribution over the sample surface. The defect map can also be used for locating the DOI for further investigation. Although the inspection tools have high throughput and are able to test several samples per hour, they are limited in resolution and are not suitable for characterizing each defect effectively. Therefore some of defects are selected for defect review and detailed characterization. In defect review, defects of interest are characterized to obtain a high resolution image and classified accurately. The defect review process relies on techniques that can provide high resolution topographical information about each defect. The examples for defect review techniques are AFM and S/TEM. Unlike inspection tools, the review techniques are relatively slower and need to accurately locate DOI for imaging.   Figure 1. a) Current process of defect inspection and review for hard disk media. The green path shows the process path for conventional AFMs. The blue path shows the process with the first generation of Park ADR-AFM. b) A defect on hard disk media sample which is marked by a surrounding scratch. The defect is hardly visible in optical vision of AFM. Locating DOI is the major challenge for defect review tools. There is a difference between the coordinates of DOI on the stage of inspection tool versus its coordinates on the stage of review tool. This difference is referred to as stage error. The stage error is generally larger for in-house inspection tools (e.g. AOI, Tester) comparing to commercial inspection tools (e.g. KLA-Tencor Candela). The process of defect inspection and review is depicted schematically in Figure 1a. In the past and before introducing ADR-AFM, manual AFM was used for imaging DOI. To facilitate locating DOI in manual AFM, FA engineers initially marked every single defect. As shown in Figure 1a hard disk media is inspected with an automated optical inspection (AOI) tool which is the most commonly used equipment in mass production and the map of defects is generated. Then the disk is investigated by an optical microscope for locating and marking each DOI. Marking defects was done by making observable scratches around each defect to facilitate finding it through the optics of AFM (Figure 1b). This process path is shown by green arrows in Figure 1a. This included performing multiple survey scans to find DOI. This method had a low throughput of 10 defects per day in best scenario. In addition, conventional AFMs were using destructive scanning method (tapping mode), therefore, tip life was limited due to performing multiple large survey scans that was required for finding each defect and the whole process resulted in high cost of ownership. The first generation of Park ADR-AFM had been introduced in order to address the limitations of conventional AFMs in the defect study process. This solution provided two major advancements comparing to conventional AFMs. First, it has a significantly improved throughput up to 10 defects per hour as a result of automated process. Since the system was operated by automated software, the user did not need to be present during the scan. Hence the system could continue locating and imaging a large number of defects on its own (even overnight). This improved the throughput of an AFM system significantly. Second, the imaging was performed in non-contact mode. Using non-contact mode imaging extended the tip life, which is critical for automated defect review process and getting consistent data. Therefore a single tip could be used for a significant number of images without user being present.   Figure 2. The AFM based defect study process is schematically shown for a) manual AFM, b) ADR-AFM. In ADR-AFM, locating the defect from defect map, survey scan, final zoom-in scan are performed automatically by the system. The ADR-AFM process is depicted in Figure 2 and compared to conventional process using manual AFM. In conventional AFM imaging (Figure 2a), the sample surface was first searched with an inspection tool. This was followed by marking the DOI on the surface using optical microscope. After laborious efforts of marking each DOI, the sample was brought to AFM for imaging. Finding the defect in AFM required visual search in vision and followed by multiple low and high magnification scans (green path in Figure 1a). In ADR-AFM, DOI coordinates from inspection tool are initially imported which is followed by coordinate alignment. After this step, automated software moves the tip to each defect, performs a survey scan, images the defect, and classifies the defect automatically. With ADR AFM throughput of defect review increases up to 100 defects per day. In addition, since this process is fully automated, engineer do not need to stand by and operate the system during the runs which frees up their time and boosts productivity. The survey scan size of old Park ADR-AFM was selected based on the stage error of the inspection tool used to generate defect maps. A 40 μm × 40 μm survey scan was typical to cover for a stage error up to 20 μm between AFM and Candela. Since the stage error of AOI tools are very large (≥ 50 μm), using a linkage tool (e.g. Candela) with smaller stager error was necessary in order to have the process of defect study complete as shown by blue path in Figure 1a. A set of 16 markers were added to each wafer before the inspection run by Candela to be used for sample alignment in Park ADR-AFM.   Figure 3. An example of enhanced vision image and how it facilitates linkage between the AOI tool which has larger stage errors and Park ADR-AFM.   The new generation of Park ADR-AFM is introduced recently for both hard disk media and 300 mm wafers [2]. Bright field enhanced vision is introduced in the new generation of Park ADR-AFM. It provides the ability to observe defects in ADR-AFM optical field of view (FOV). Bright field enhanced vision utilizes differential frame averaging technique and precise movement of sample by decoupled XY scanner to observe small defects that are hardly visible through regular optics of the system. Decoupled XY and Z scanner structure of Park ADR-AFM enables using differential frame averaging by collecting optical images of the sample surface at two precisely separated positions of XY scanner. The differential image of the frames of sample surface collected at two different positions enhances the contrast of small defects and makes them visible in optics of ADR-AFM. Since the FOV of ADR-AFM is larger than the maximum survey scan size, enhanced vision enables the ADR-AFM to accommodate even larger stage errors of inspection tools (Figure 3). Collecting a typical bright field enhanced vision image takes between 1 – 30 sec for each defect.   Figure 4. The schematic shows the sample alignment process in the first (left) versus the new (right) generation of Park ADR-AFM. In the latest Park ADR-AFM only one marker is needed to indicate zero-degree location of hard disk media samples.   In addition to bright field enhanced vision, a new sample alignment algorithm is developed to facilitate the linkage between the ADR-AFM and inspection tools. In the first generation of ADR-AFM, a sample alignment by using 16 markers was needed to align the sample on the ADR-AFM stage. The coordinates of 16 markers were generated by Candela, and were matched with the coordinates of ADR-AFM stage. In the next generation of ADR-AFM, only one marker is needed for sample alignment. The new alignment process consists of one coarse and one fine alignment step. In the coarse alignment, three points at the circular perimeter of the hard disk media sample is used for finding the sample center, and a zero-degree marker is used for finding the rotation of the sample. In the fine alignment step, ADR-AFM uses few larger defects for minimizing the remaining positioning errors by utilizing enhanced vision. The new alignment process is faster and more convenient for sample alignment comparing to the first generation of ADRAFM. In addition, it can also be used as a tool for evaluating the stage error of AOI or other inspection tools (Figure 4).   Figure 5. The new ADR-AFM process with newly introduced dark field enhanced vision for hard disk media with reflective surface. Dark field enhanced vision is introduced in the latest generation of Park ADR-AFM. The new dark field enhanced vision enables AFM to observe defects in a short time on reflective surfaces by collecting the light scattered due to surface defects. For this purpose, a new hardware module is developed and implemented in the AFM system. In this module a laser beam is emitted to the location of interest on the sample. The beam angle of incidence is larger than zero degrees such that the direct beam reflection is not collected by the on-axis optics of ADR-AFM. For a reflective sample, majority of laser beam is following the expected reflection path with similar angle of incidence. In case the surface a topological defect is present on the sample surface, part of the laser beam will be scattered in different directions other than the regular reflection path and are collected by on-axis ADR-AFM optics. The collected frame is an image of the surface with collected scattered light from the defects. Examples of standard optical and dark field enhanced vision images are shown in Figure 5. The upgraded ADR-AFM process with dark field enhanced vision is presented schematically in Figure 5. In the new process, after the sample alignment, ADR-AFM goes to defect location, recognizes the defect using dark field enhanced vision (without AFM scanning), and performs the imaging, analysis, and classification. The new process enables ADR-AFM to minimize the positioning error. As a result the required survey scan is also minimized and the throughput is increased. As a result, it enables ADR-AFM to accommodate defect maps from AOI systems that have large stage errors. The ability to see the defects by enhanced vision enables ADR-AFM to be used as the reference tool for evaluating the stage errors of different inspection tools.   Figure 6. The results of performing ADR-AFM on a 95 mm hard disk media samples for a set of defects reported by an in-house inspection tool. 33 defects out of 37 was found and imaged by ADR-AFM. The results of a defect review by ADR-AFM on a 95 mm hard disk media sample are shown in Figure 6. The defect coordinates were provided by an in-house inspection tool with large stage error (> 50 μm). Dark field enhanced vision has been used for locating the defects. As the result, ADR-AFM was able to find 33 out of 37 defects successfully using the addition of enhanced vision. The defects are classified to two groups of pit and bump by ADR-AFM. An example of each group is depicted in Figure 7. In this figure dark field enhanced vision images and AFM images of defects are shown. ADR-AFM was able to find defects with vertical dimension as low as < 3nm.   Figure 7. Representative pit and bump type defects on a hard disk media samples which were found and imaged by ADR-AFM. The dark field enhanced vision images are shown side by side the AFM images and selected profiles of the defects. In summary, the latest generation of Park ADR-AFM is equipped with new bright field and dark field enhanced vision modules. The bright field enhanced vision utilizes frame averaging technique and ADR-AFM’s decoupled XY scanner to obtain a high resolution optical image of the sample surface to help locating the DOI. Dark field enhanced vision is enabled by collecting the scattered laser signal by the on-axis vision of ADR-AFM and enables detecting defects on the surface of reflective surfaces. Park ADRAFM uses non-contact mode imaging and therefore is imaging the samples in a non-destructive method. Park ADRAFM with enhanced vision modules and updated alignment algorithms is used for high throughput AFM based defect review. Additionally, ADR-AFM with enhanced vision is a potential reference tool for evaluating the stage error for inhouse inspection systems that are being developed or maintained.   REFERENCES [1] Smith, G. T., Industrial Metrology: Surfaces and Roundness, Springer, London, 103-105 (2002). [2] Ardavan Zandiatashbar et al., "Highthroughput automatic defect review for 300mm blank wafers with atomic force microscope," in Proc. SPIE 9424, Metrology, Inspection, and Process Control for Microlithography XXIX, 2015, p. 94241X.

8 Jan 2015

Sub-Angstrom Roughness Repeatability With Tip-to-tip Correlation

By Ardavan Zandiatashbar, Park Systems Inc., Santa Clara, California. ABSTRACT Nano-scale surface roughness measurement has been one of the most common applications of atomic force microscopes over the past two decades. Roughness measurement accuracy and repeatability along with tip life are the major factors that had been considered to be almost impossible to coexist. In this article, we review the basics of non-contact (attractive) mode imaging and introduce soft dynamic mode for special sub-angstrom level roughness measurement applications. The effects of using soft dynamic mode on tip life, measurement consistency and tip-to-tip correlation have been investigated. It is found that soft dynamic mode could be the solution for samples with sub-angstrom surface roughness covered with liquid layer.   INTRODUCTION Characterization of nano-scale surface roughness and statistical properties of surfaces for different materials is known to be one of the most common applications of scanning probe microscopy (SPM) both in research and production. As an example for SPM application in industrial research and development, surface roughness measurement is required for quality control of glass media samples or heat-assisted magnetic recording (HAMR) products in data storage industry. On the other hand, surface quality control after chemomechanical polishing (CMP) is key factor in determining the reliability of final products in semiconductor production. Nanometer scale roughness of production wafers is important for device patterning in integrated circuit (IC) manufacturing. In all of these applications, understanding the physical property of the sample surface at nano-scale is essential to maintain the reliability and understand failure mechanisms in final products. Atomic force microscopy (AFM) plays a key role in nano-scale surface roughness measurement due to its sub-nanometer level resolution in Z direction. Non-contact (NC) mode or attractive state oscillation imaging has been essential for preserving tip life and consistency in measurements. Although NC mode was the first dynamic force microscopy mode [1], lack of high bandwidth Z-actuator made it difficult to maintain the tip consistently in attractive state. Therefore repulsive state oscillation was used as an alternative which is also known as dynamic mode. Dynamic mode is usually used with high oscillation amplitudes, which leads to limited tip lifetime. Consequently, the data measurement lacks repeatability as it relies heavily on the tip sharpness. Using blunt tips affects significantly the surface roughness measurements as the surface feature size become comparable to or smaller than effective tip radius. Park Systems pioneered manufacturing commercial AFM systems with the focus on performing non-contact mode imaging with high bandwidth Z-actuators and decoupled Z and XY scanners. Here we explain the advances made in AFM imaging for subangstrom surface roughness measurement and repeatability using Park NX system.   DISCUSSIONS NON-CONTACT VS. DYNAMIC MODE To perform amplitude modulation non-contact mode imaging, the frequency of the cantilever oscillation should be selected slightly larger than the resonance frequency of the cantilever. This frequency selection is based on the fact that the cantilever resonance frequency increases when the cantilever oscillates in attractive state based on the tip-sample separation. In this state, the phase of oscillation for a sharp clean tip is negative in the AFM systems manufactured by Park Systems. (The actual value and sign of the phase signal varies depending on system manufacturer). While Z stage is approaching the sample surface, the cantilever oscillation amplitude is affected by tip-sample interaction forces. This behavior can be observed in Figure 1. In this figure an experimental amplitudedisplacement (A-d) curve is shown. In this curve the oscillation amplitude is measured as the tip approaches towards and then retracts from the sample surface. While the tip-sample separation decreases (red curve), the tipsample interaction is in attractive state (NC mode) until a jump in amplitude occurs and tip-sample interaction transitions to repulsive state (dynamic mode). Similarly when the tip is retracting from the surface (blue curve), it suddenly switches from repulsive to attractive state. As a rule of thumb, this transition generally happens when the tip-sample separation is almost between half to three quarters of free oscillation amplitude of cantilever. This transitional behavior between NC and dynamic modes can be seen as a hysteresis loop in the plot and creates a bistable state [2]. In this state, the feedback system becomes bistable. A corresponding hysteresis loop can be observed in the phase curve as well. In the phase hysteresis loop, phase values transition between negative (attractive state or NC mode) and positive values (repulsive state or dynamic mode). Figure 1. AFM cantilever oscillation amplitude (upper) and phase (lower) vs. tip-sample separation (d) known as A-d curve for constant Z actuator driving power.   TIP SHARPNESS BASED ON A-D CURVES The amplitude curves can be affected by tip sharpness as well. Figure 2 shows three sets of amplitude and phase curves representing sharp, normal, and blunt tips. The tip sharpness is verified by scanning a 70 nm grating. As the tip loses its sharpness, its ability to precisely resolve the gratings becomes limited, while the attractive state region becomes slightly wider. For a normal tip it is easier to maintain the tip in attractive state as this state is wider. For a blunt tip, the amplitude and phase curves do not include the bistable state hysteresis loop, and phase can reach large negative values. In fact, Figure 2 shows that amplitude curve can be used as an estimated tip sharpness/quality characterization tool. Plots on the left show representative A-d curves (plots of cantilever oscillation amplitude and phase vs. tip sample separation) for sharp, typical, and blunt tips. The plots on the right side are the corresponding profiles of a calibration grid collected with each tip.   DYNAMIC MODE BENEFITS AND LIMITATIONS NC mode imaging has the benefits of preserving tip life and measurement consistency, and has been very helpful in measurement of sub-angstrom level roughness values, for example for bare silicon wafers. However, in some cases its performance may be limited when measuring samples with small roughness values (<2Å). In these cases, the image loses resolution and has poor quality. This could be attributed to the presence of liquid layer on top of the sample surface. If the thickness of the liquid layer is comparable to the feature size, it does not allow the tip to track the surface features accurately in attractive state. In order to overcome this issue, tip needs to oscillate in repulsive state. This way it can penetrate the liquid layer and reach the sample surface (contact). This phenomenon is demonstrated by imaging a hard drive glass media sample using both NC and dynamic modes in Figure 3. In this image the sample was initially imaged using the NC mode with the cantilever oscillation amplitude of 10 nm. Despite the acceptable sharpness of the tip, the image quality is not desirable. Sample was imaged again using dynamic mode (with oscillation frequency smaller than the free air resonance frequency (f0)). The dynamic mode image has better quality than the initial image. In the 3rd scan, the sample was imaged in NC mode by using cantilever oscillation frequency larger than f0 and inability of the tip reaching the sample surface was confirmed. Figure 3. Imaging sample surface with angstrom level roughness, covered with liquid layer collected in tip-sample interaction being a) attractive, b) repulsive, & c) attractive again.   SOFT DYNAMIC MODE Conventionally dynamic mode imaging is performed with large amplitude oscillation (>20 nm). The larger amplitude oscillation is enabled by applying more energy to the cantilever which facilitates the tip ability to “contact” the sample surface. In addition, the AFM systems ability to perform dynamic mode imaging with low oscillation amplitude was limited due to the bandwidth of the Z servo of the AFM system and system noise level. If a system can perform dynamic mode with low amplitude, it should be able to apply minimum energy to the cantilever, just enough to penetrate the liquid layer, then the tip life may be preserved longer and minimize its effect on the measurement repeatability. Park NX series were designed for low amplitude NC mode imaging, hence it is made possible to perform low amplitude low energy cantilever oscillation with maximum signal to noise ratio. We call this type of dynamic mode (with low oscillation amplitude) soft dynamic (SD) mode. Figure 4 shows a comparison between the tip life in normal dynamic mode (high amplitude) vs. SD mode (low amplitude) by scanning a glass substrate sample with roughness value of about 1Å. In this figure, the 1st and 50th scanned images by normal dynamic mode are compared. The 1um x 0.25um images (512 x 128 pixels) are collected with scan rate of 1Hz. The roughness value (Ra) is decreased by 6% over 50 images and image resolution significantly degrades due to loss of tip sharpness. The two lower images show the comparison between 1st and 1000th images collected with same tip using SD mode. The decrease in roughness value is below 2% over 1000 images, and high image resolution is well-preserved until the last image. Performing SD mode imaging is made possible by using fast Z servo feedback. Despite the gradual tip wear, SD mode clearly outperforms normal dynamic mode in terms of tip life and measurement consistency. Figure 4. Repeatability test using conventional (upper) and low amplitude soft (lower) dynamic mode on glass substrate.   TIP TO TIP VARIATION AND TIP LIFE Another important factor about measurement consistency is tip-to-tip correlation. It is desirable to eliminate tip dependency of the measurements in order to obtain true sample surface property by AFM roughness measurement. For this purpose, the 1000-image roughness measurement test was repeated with three different tips. Variations of roughness (Ra) over 1000 scans for the three tips are shown in Figure 5. The average and standard deviation of the data set for each tip is shown in the inset table. It can be seen that for the sample with Ra of 1Å, the repeatability measurements show variations below 0.03Å (below 3%). This minimal variation shows tip independency of roughness measurements using SD mode. Figure 5. 1000 repeat roughness measurement variation   SAMPLE VARIATIONS Although the measurement variation is small (<3%), the source of the variation still remains unknown. Figure 6 shows three sets of roughness measurement on different positions on the HDD glass media. At each location, 1um x 2um images with resolution of 512 x 1024 pixels are collected at scan rate of 1Hz in SD mode. Each image is divided to eight 512 x 128 pixel sections in order to be comparable with images collected for Figure 5 and Ra values were measured for each subimage. It turned out that the Ra variation on the sample surface from location to location is around 3% which is comparable to the variation observed in Figure 5. Therefore, one possible source of the 3% variation observed in Figure 5 could be attributed to point to point variation on the sample surface. Figure 6. Roughness measurement images at different locations on sample surface. The roughness values show around 3% variation from location to location in 1 μm x 0.25 μm.   CONCLUSION In this article we introduced SD mode imaging as an alternative method for subangstrom surface roughness measurement by using low amplitude low energy cantilever oscillation with Park NX series. We performed 1000 roughness measurements using one single tip using SD mode and showed less than 3% variation while the image resolution and tip sharpness is preserved. This variation is identical to the variation of roughness over different locations on the sample surface. By repeating the 1000 scan test using two additional tips, significant tip-to-tip correlation with 3% variation was shown to exist. Performing SD mode imaging is made possible by using large bandwidth and high speed feedback system provided on Park NX series. This solution is provided for research and production in different industries including hard drive media and semiconductor manufacturing. REFERENCES [1] G. Binnig, C. F. Quate and C. Gerber, "Atomic Force Microscope," Physical Review Letters, pp. 930-933, 1986. [2] R. Garcia and R. Perez, "Dynamic atomic force microscopy methods," Surface Science Reports, pp. 197-301, 2002.

8 Dec 2014

Automated AFM Significantly Boosts Throughput in Automatic Defect Review

Reduces Cost and Increases Efficiency in Production of Semiconductor Wafers   Introduction The scaling trend in semiconductor design requires tighter control over defects on wafers. In order to characterize the defects, both inspection and review tools are needed. The inspection tools such as surface scanning inspection systems (SSIS) are employed first to locate and map defect sites, and the review tools such as scanning electron microscopy (SEM) are used to obtain morphological information. Large defects, which are visible by optical microscopy, can be easily found by an inspection tool and imaged by a review tool. However, as processing requirements become more stringent, the defect size of interest gets smaller, below the diffraction limit of optical microscopy. To locate and identify small defects, a large survey scan needs to be performed by the review tool in order to find the exact location of the defects. Figure 1. The goal of defect review is to locate the defect site identified by an inspection tool and image the defect by a review tool Defect review SEM has been used to image the defects of a bare wafer after defect sites are mapped by the laser-scattering defect inspection tools such as Surfscan from KLATencor. However, the imaging by defect review SEM is limited to 2D and cannot provide the 3D information, which has become more crucial in recent days for bare wafer manufacturing control. Also, smaller defects, less than 50nm, are increasingly not detected by defect review SEM. Why Automatic Defect Review (ADR) AFM? Atomic force microscopy (AFM) is gaining more importance as the choice of a review tool in recent days because it can directly measure the defect dimensions (height and width, hence shape) and the physical properties of defects of interest. However, for engineers and researchers working with bare wafers, the process of identifying nanoscale defects is time consuming with conventional manual AFM, hindering productivity, let alone throughput. Thus it is important to have an automated solution for both failure analysis and production engineers who need to find the cause of defect sites on bare wafers and minimize them. Figure 2. Typical routine of defect review process by manual AFM has a very low throughput, 10 defects per day at best.  Moreover, the tip cost can run high if one uses a destructive scan mode such as tapping. To address this need, Park Systems have delivered the Automatic Defect Review (ADR) AFM which speeds up and improves the way defects are imaged and analyzed. With this solution, users are now able to obtain the additional details and the height/depth information of the defects which were not possible or too expensive to obtain with SEM based ADR. How to Transfer Defect Map from Inspection Tool to Review AFM The general defect review concept is simple, yet complicated when it is done on the nanometer scale. Simple, because all it would require is the proper transferring of the defect map from an inspection tool and re-locating those same defects on a review tool. Coordinates from the inspection tool are translated to the review tool, which is used to zoom in and image the details of the defect, revealing its characteristics. The complicated part comes into play when attempting to accurately go near the defect site, in order to image it, and to do this all in automation. This is due to the fact that there are stage mapping errors between the inspection tool and the review tool. In case of a patterned wafer, alignment marks are used as reference points to map the two stages between the inspection tool and AFM, and the position error from both tools can be accurately estimated. Park’s initial success in Automated Defect Review (ADR) AFM solution was implemented in the hard disk industry, utilizing the reference marks inscribed by the optical inspection tool such as KLA Tencor’s Candela series. Utilizing these marks, the defect maps of hard disk media or substrates could be accurately and reliably transferred and the automated review AFM, Park HDM Series, was able to get near enough to the defect to conduct a survey scan, and then follow up with a zoom-in scan to provide the details the user required. As a whole, this solution proved to be effective, increasing the productivity in the HDD industry’s FA labs by up to 1000%. Automated, Accurate Transfer of Defect Map WITHOUT Reference Marks Success in the HDD industry resulted in similar requests coming from the semiconductor industry. Since the cleanness level requirement for the semiconductor industry is much stricter than that of HDD, creating markers on the bare wafers is not possible, hence the need for ADR process WITHOUT any reference marking on a bare wafer is required. Without alignment marks the positions of defects are roughly estimated [1, 2]. Therefore, additional alignment process is required to decrease the positional error [3] during the defect map transfer. To address this issue, a new method was developed that utilizes the wafer edge, notch, and large defects, which are visible by optical microscopy, as the reference points. Figure 3. Wafer edge and actual defects are used to orient the wafer and translate the coordinate between the inspection tool and Park AFM.    The improved remapping technique, combined with enhanced vision, does not require any reference marker on a bare wafer. Furthermore, it also does not need any separate step to calibrate the stage of the targeted inspection system. The success rate of the 300mm bare wafer ADR depends on two factors: the accuracy of stage mapping and the size of initial survey scan. The more accurate the stage mapping, the smaller survey scan size can be used for higher success in finding the defect. The new stage mapping technique vastly improves the accuracy by adopting advanced remapping algorithm and enhanced vision technique. The RMS position error of survey scans are less than 5 μm regardless of wafer loading positions. It means that all the defects are located within ±5 μm using the defect map provided from a defect inspection tool. Figure 4. The positional error of actual defect sites after the automatic stage mapping is quite small, about 5µm in rms. The positional error remains largely the same regardless of wafer loading positions.  Conclusion: Results of 300 mm Bare Wafers ADR by Park NX-Wafer Utilizing the advanced coordinate translation technique, the defect map obtained from a laser-scattering defect inspection tool, such as Surfscan from KLA-Tencor, can be accurately transferred to a 300mm AFM system, allowing full automation for high throughput defect imaging. With the new 300mm bare wafer ADR, a typical automated measurement run would go as follows. - A bare wafer is run by a laser-scattering defect inspection tool (e.g. Surfscan) - Then an operator registers the resulting defect coordinates file, known as KLARF for the Surfscan tool, to Park NX-Wafer. - The coordinates of the defect map are automatically de-skewed and the linkage of the two stages is enabled between the inspection tool and Park NX-Wafer - Park NX-Wafer then runs the automatic defect review (ADR) on the sample bare wafer. The above set of data shows the results from the 300mm bare wafer ADR with Park NX-Wafer. Typical success rate is higher than 95% regardless of the orientation of a wafer. The zoomed-in scans show the detailed morphology of defects, clearly distinguishing various types such as pit, bump, and scratch. Figure 5. The 300mm bare wafer ADR AFM correctly locates all the defect sites identified from an inspection tool and automatically zooms in for the detailed imaging.   Figure 5 (a)   Figure 5 (b)   Figure 5 (c)    Figure 6. The AFM imaging shows the residual damage left during the SEM imaging.

1 Jul 2014

Automatic Defect Review (ADR) of 300mm Bare Wafers with Automated AFM

The scaling trend in semiconductor design requires tighter control over defects on wafers. In order to characterize the defects, both inspection and review tools are needed.  The inspection tools such as Surface Scanning Inspection Systems (SSIS) are employed first to locate and map defect sites, and the review tools such as Scanning Electron Microscopy (SEM) are used to obtain morphological information. Large defects, which are visible by optical microscopy, can be easily found by an inspection tool and imaged by a review tool. However, as processing requirements become more stringent, the defect size of interest gets smaller, below the diffraction limit of optical microscopy. To locate and identify small defects, a large survey scan needs to be performed by the review tool in order to find the exact location of the defects. Figure 1. The goal of defect review is to locate the defect site identified by an inspection tool and image the defect by a review tool Defect review Scanning Electron Microscope (SEM) has been used to image the defects of a bare wafer after defect sites are mapped by the laser-scattering defect inspection tools such as Surfscan from KLA-Tencor. However, the imaging by defect review SEM is limited to 2D and cannot provide the 3D information, which has become more crucial in recent days for bare wafer manufacturing control. Also, smaller defects, less than 50nm, are increasingly not detected by defect review SEM. Why Automatic Defect Review AFM? Atomic Force Microscopy (AFM) is gaining more importance as the choice of a review tool in recent days because it can directly measure the defect dimensions (height and width, hence shape) and the physical properties of defects of interest. However, for engineers and researchers working with bare wafers, the process of identifying nanoscale defects is time consuming with conventional manual AFM, hindering productivity, let alone throughput.  Thus it is important to have an automated solution for both failure analysis and production engineers who need to find the cause of defect sites on bare wafers and minimize them.   Figure 2. Typical routine of defect review process by manual AFM has a very low throughput, 10 defects per day at best.  Moreover, the tip cost can run high if one uses a destructive scan mode such as tapping. To address this need, Park Systems have delivered the Automatic Defect Review (ADR) AFM which speeds up and improves the way defects are imaged and analyzed.  With this solution, users are now able to obtain the additional details and the height/depth information of the defects which were not possible or too expensive to obtain with SEM based ADR. How to Transfer Defect Map from Inspection Tool to Review AFM The general defect review concept is simple, yet complicated when it is done on the nanometer scale.  Simple, because all it would require is the proper transferring of the defect map from an inspection tool and re-locating those same defects on a review tool.  Coordinates from the inspection tool are translated to the review tool, which is used to zoom in and image the details of the defect, revealing its characteristics.  The complicated part comes into play when attempting to accurately go near enough to the defect site, in order to image it, and to do this all in automation.  When the two stages of an inspection tool and review tool are linked, there are various errors to be accounted for.  The successful outcome of the automatic defect review process depends on the accuracy of the two stage mapping. In case of a patterned wafer, alignment marks are used as reference points to map the two stages between the inspection tool and AFM, and the position error from both tools can be accurately estimated. Park Systems’ initial success in Automated Defect Review (ADR) AFM solution was implemented in the hard disk industry, utilizing the reference marks inscribed by the optical inspection tool such as KLA Tencor’s Candela series.  Utilizing these marks, the defect maps of hard disk media or substrates could be accurately and reliably transferred and the automated review AFM, Park HDM Series, was able to get near enough to the defect to conduct a survey scan, and then follow up with a zoom-in scan to provide the details the user required.  As a whole, this solution proved to be effective, increasing the productivity in the HDD industry’s FA labs by up to 1000%. Automated, Accurate Transfer of Defect Map WITHOUT Reference Marks Success in the HDD industry resulted in similar requests coming from the semiconductor industry as well.  Since the cleanness level requirement for the semiconductor industry is much stricter than HDD, creating markers on the bare wafers was not possible, hence the need for ADR process WITHOUT any reference marking on a bare wafer.  Without alignment marks the positions of defects are roughly estimated [1, 2].  Hence, additional alignment process is required to decrease the positional error [3] during the defect map transfer.  To address this issue, a new method was developed that utilizes the wafer edge, notch, and large defects, which are visible by optical microscopy, as the reference points. The improved remapping technique, combined with enhanced vision, does not require any reference marker on a bare wafer.  It also does not need any separate step to calibrate the stage of the targeted inspection system. Figure 3. Wafer edge and actual defects are used to orient the wafer and translate the coordinate between the inspection tool and Park AFM.  The success rate of the 300mm bare wafer ADR depends on two factors: the accuracy of stage mapping and the size of initial survey scan.  The more accurate the stage mapping is, the smaller survey scan size can be used, hence the higher success of defect finding.  The new stage mapping technique vastly improved the accuracy by adopting advanced remapping algorithm and enhanced vision technique.  The RMS position error of survey scans are less than 5 µm regardless of wafer loading positions.  It means that all the defects are located within ±5 µm using the defect map provided from a defect inspection tool    Figure 4. The positional error of actual defect sites after the automatic stage mapping is quite small, about 5µm in rms. The positional error remains largely the same regardless of wafer loading positions.  Results of 300mm Bare Wafer ADR by Park NX-Wafer Utilizing the advanced coordinate translation technique, the defect map obtained from a laser-scattering defect inspection tool, such as Surfscan from KLA-Tencor, can be accurately transferred to a 300mm AFM system, allowing full automation for high throughput defect imaging. With the new 300mm bare wafer ADR, a typical automated measurement run would go as follows. - A bare wafer is run by a laser-scattering defect inspection tool (e.g. Surfscan) - Then an operator registers the resulting defect coordinates file, known as KLARF for the Surfscan tool, to Park NX-Wafer. - The coordinates of the defect map are automatically de-skewed and the linkage of the two stages is enabled between the inspection tool and Park NX-Wafer - Park NX-Wafer then runs the automatic defect review (ADR) on the sample bare wafer. The following set of data shows the results from the 300mm bare wafer ADR with Park NX-Wafer.  Typical success rate is higher than 95% regardless of the orientation of a wafer.  The zoomed-in scans show the detailed morphology of defects, clearly distinguishing various types such as pit, bump, and scratch.   Figure 5. The 300mm bare wafer ADR AFM correctly locates all the defect sites identified from an inspection tool and automatically zooms in for the detailed imaging. Figure 5 (a)   Figure 5 (b) Figure 5 (c)     It is notable that defect review SEM leaves destructive irradiation marks, square-shaped, on defect sites after its run.  During the ADR run of some wafer samples, the survey scan and zoom-in scan found the residual damage left on defect sites by the SEM imaging. The irradiation damage during SEM review process could limit its use in production control.   Figure 6. The AFM imaging shows the residual damage left during the SEM imaging. In order for automated ADR to work effectively, two important criteria must be met: ultra low-noise system for small sized defects and tip preservation for automated process without interruption.  Delivering the industry’s best low-noise system, at less than 0.5Å RMS for a large sample AFM system, plus its True Non-Contact measurement mode, Park Systems’ makes available a powerful and working AFM solution for automatic defect review process.  References Ota, H., Hachiya, M., Ichiyasu, Y., & Kurenuma, T., Hitachi Review, 55(2), 79 (2006). Nutsch, A., Funakoshi, T., Pfitzner, L., Steffen, R., Supplieth, F., & Ryssel, H., ISSM 2007. pp. 1-4 (2007). Kinney, P. D., & Uritsky, Y. S., International Society for Optics and Photonics. In Microelectronic Manufacturing pp. 204-209. (1998)

27 Dec 2013

Undercut Structures and Sidewall Roughness

XE-3DM The XE-3DM metrology AFM incorporates the decoupled scanner configuration.   Because the Z scanner is independent of the XY scanner, it can be intentionally tilted with respect to the XY plane. By this mean, the cantilever tip can easily reach the sidewall and undercut structure.   Figure 1. A undercut structure imaged by XE-3DM, showing standing wave at the side wall   Except for the tilting Z scanner, this new 3D AFM operates in the same way as regular AFM, so it can take advantage of standard super sharp cantilever tips, instead of blunt flared tips for imaging. This makes it an ideal system for high resolution sidewall roughness measurements with significantly reduced blind spots near the bottom corner areas.     Figure 2. Photoresist sidewall roughness measurement. (A), A 3D AFM image of a photoresist sample showing three different region: top surface, sidewall, and bottom surface; (B) measured roughness versus the position cross the sample surface showing different surface roughness for different region

26 Dec 2013

Automatic Defect Review AFM for Hard Disk Media and Substrates

Park Systems, Nanotechnology Solutions Partner for HDD Industry Park Systems serves the hard disk drive (HDD) industry with automated nanotechnology measurement solutions including atomic force microscopes, software, and global service and support. Partnering with world leaders in HDD industry, Park Systems has been successfully delivering optimized solutions for the most challenging imaging and measurement needs in the industry.  The company's HDD-Optimize program is the state-of the-art nanotechnology solution service for the hard disk drive industry. Under the systematic approach of the program, Park Systems ensures rapid alignment and performance of the Park AFM nanotechnology measurement platform with the specific requirement of its customers. Limitations in Current Method of HDM Defect Review   Defect Mapping by Tester Defect location/marking with KLA Tencor’s CandelaTM Throughput: less than 5 defects/hr Manual AFM imaging Throughput: : 1~2 defects/hr  Low throughput: 10 defects per day at best  High labor cost  Destructive method Challenges in Developing Automated Defect Review AFM During Survey Scan;  Small defects are imaged in one or two pixels.  Signal level is often low                 We need AFM with;   Low system noise, less than 0.5 A rms  Artifact-free AFM scan  - to detect tiny defect signal during the Survey Scan  - to profile the defect type (pit, bump, scratch, etc) XE-HDM: Enabling Automatic Defect Review of Hard Disk Media and Substrates Continuing the company’s impressive track record of developing optimized solutions, Park Systems introduces XE-HDM, an automatic defect review AFM which revolutionizes the way defects in HDD substrates and media are searched, scanned and analyzed. The new XE-HDM significantly increases throughput for the defect review process.  Automated reference marker detection  Automated stage mapping of XE-HDM  Transfer KLA Tencor CandelaTM defect map to XE-HDM  Automated survey scan of defects mapped by CandelaTM  Automated zoom-in scan of specified defects  Automated profiling of imaged defect types  Automated analysis of imaged defects (in development) The Limitations of Conventional AFM Scanner Bowing (Piezo tube is NOT an Orthogonal 3-D Actuator) Flattening, software correction, is required Distorts or wipes out tiny defect signals during the Survey Scan Non-Contact Mode not Possible Inevitable tip wear occurs when using contact or tapping mode Tip life is not long enough to complete “one” automated defect review run The Innovative Park AFM Technology Overcomes the Challenges Flat and Linear XY Scan (Decoupled XY and Z Scanners) No Flattening is required, providing flat and linear XY scan All defects can be detected during the Survey Scan without any artifacts True Non-Contact Mode (High Force Z Scanner) Significantly increased tip life (less tip wear and tear) Consistent tip condition throughout multiple automatic runs Result of Test Runs The test runs demonstrate over 500 ~ 800 % gain in throughput The automation allows users to manage several tools simultaneously Summary: Key Benefits of XE-HDM

26 Dec 2013

Surface Roughness Measurement of Media and Substrate

Surface Roughness Measurement of Media and Substrate Eunji Shin (Industrial Product Management of Park Systems, Seoul, Korea)   Atomic force microscopy (AFM) is useful for investigating the surfaces and characteristics of different materials down to nanometer detail1-6.  Accurate surface roughness measurement is not only a highly powerful technique for understanding the basic physics of materials, but also very useful in examining device structures and their failure mechanisms. In addition, roughness repeatability is one of the very important factors in industrial applications as it relates to the customer’s product reliability. Park AFM’s True Non-Contact Mode is especially essential for accurate and repeatable surface roughness measurement. In order to measure accurate and repeatable roughness, two key requirements need to be satisfied. First is maintaining the tip sharpness and the other is lowering the system noise. In this paper, we examine these two requirements for measuring surface roughness. Also, we demonstrate why True Non-Contact mode is powerful for accurate roughness measurement by testing a tip-check sample and ultra flat media samples using Park AFM.   Requirements for Accurate Roughness Measurement To measure accurate and repeatable roughness, the first thing to consider is the tip effect. That is because the tip geometry directly correlates to the roughness value. As shown in Figure 1, when the tip is sharp, it obtains an accurate feature of the surface. But as the tip becomes blunt, it cannot approach the bottom of the sample, so it obtains a skewed feature of the surface. In short, as the tip radius increases, the roughness value decreases. That is why the tip sharpness needs to be preserved during the roughness measurements. Figure 1. (a), (b) show the tip effect on roughness measurement.   The system noise is equally important. The reason is when the noise gets mixed into the AFM results, the AFM cannot accurately measure the ultra flat surface. To detect the smallest of sample features and to image the flattest of surfaces, low system noise is required. Therefore, Park Systems has engineered instruments that hold the industry’s lowest noise floor specification of < 0.5Å. As shown in Figure 2a, Park AFM systems routinely achieves about 0.2Å rms. Hence, it is good enough to measure the ultra flat sample, as shown in Figure 2b.  Figure 2. (a) noise floor data is determined using a “zero scan”. With the cantilever in contact with the sample surface, the system noise is measured at a single point under the following conditions: 0 x 0 nm scan, staying in one point / 0.5 gain, in contact mode / 256 x 256 pixels. (b) The roughness image of ultra flat surface was acquired with an XE-HDM in True Non-Contact mode at 1 Hz with 256 pixel resolution. Accurate and Repeatable Surface Roughness Measurement by True Non-Contact Mode In Figure 3, we evaluated how the tapping and the non-contact modes affect tip properties over sequential scans by imaging a tip tester sample of CrN layer (Swiss Center for Electronics and Microtechnology, Inc., SA). As shown in Figure 3a, the non-contact mode obtained consistent and sharp image quality even after the 100th scan. But in the tapping mode, the tip wore down quickly even before the 10th scan affecting the scan image quality significantly; the tip sharpness degraded more than 80% as shown in Figure 3b. On the other hand, True Non-Contact mode preserved the tip sharpness even after imaging 200 images of CrN. Whereas in the tapping mode, the tip diameter changed significantly as its tip end got damaged progressively during the scans. Figure 3. AFM Measurements of CrN Tip Tester Sample in different Imaging Modes. (a) True Non-Contact mode images from 100 sequential scans and the tapping mode images from 10 sequential scans. (b) The ETD(@5nm) on tip check between the True Non-Contact mode and  Peak Force Tapping mode.   The data results in Figure 4 show that when tip sharpness is preserved, the accurate roughness measurement is also preserved. We took more than 700 images in non-contact mode of the same location on the sample with one tip, and monitored a long term variation in surface roughness measurement. We acquired the average roughness value of 1.498Å (Ra) with a standard deviation (σ) of 0.024Å as shown in Figure 4a. In Figure 4b, we can observe that the high resolution imaging was maintained during the 752 sequential imaging. Figure 4c shows the preserved tip profile that was obtained by VLSI CD Standard sample. The tip shape before and after all the measurements are virtually the same, which indicates there were no noticeable tip wear during the measurements.  Figure 4. AFM measurements of hard disk drive media sample in Non-Contact Mode. (a) It shows the consistent and reliable data throughout 752 repeated measurements. (b) Non-Contact mode images from 752 sequential measurements. (c) The tip profile measured with VLSI CD Standard sample, both before and after 752 repeats.   Hence, we have shown the reliability of the non-contact AFM for surface roughness measurement and topographical imaging with sub-angstrom precision. In Figure 1 and 2, we have demonstrated why tip sharpness and low system noise are required for surface roughness measurement. Also, the AFM images and the ETD analysis in Figures 3 and 4 show that only True Non-Contact mode minimizes tip degradation, keeps sample free from damage, and prolongs high-resolution imaging throughout the AFM scan. Hence, it is clear that the non-contact mode AFM has unique characterization capabilities for industrial applications.   Reference 1G. Binning, C. Quate, and C. Gerber, Phys. Rev. Lett. 56, 930 (1986) 2D. R. Baselt and J. D. Baldeschwieler, Rev. Sci. Instrum. 64, 908 (1993). 3P. K. Hansma, B. Drake, D. Grigg, C. B. Prater, F. Yashar, G. Gurley, V. Eligns, S. Feinstein, and R. Lal, J. Appl. Phys. 76, 796 (1994) 4R. Barrett, Rev. Sci. Instrum. 62, 1393 (1991). 5K. Nakano, Rev. Sci. Instrum. 69, 1406 (1998). 6J. Kwon, J. Hong, Y. S. Kim, D. Y. Lee, K. Lee, S. M. Lee, and S. I. Park, Rev. Sci. Instrum. 74, 4378 (2003)

24 Dec 2013

High Throughput and Non-Destructive Sidewall Roughness Measurement Using 3-Dimensional AFM

High Throughput and Non-Destructive Sidewall Roughness Measurement Using 3-Dimensional Atomic Force Microscopy Yueming Hua, Cynthia Buenviaje-Coggins (Park Systems Inc., CA, USA) Yong-ha Lee, Sang-il Park (Park Systems Corp., Suwon, Korea)   ABSTRACT As the feature size of the semiconductor device is becoming increasingly smaller and the transistor has become three-dimensional (e.g. Fin-FET structure), a simple Line Edge Roughness (LER) is no longer sufficient for characterizing these devices. Sidewall Roughness (SWR) is now the more proper metric for these metrology applications. However, current metrology technologies, such as SEM and OCD, provide limited information on the sidewall of such small structures. The subject of this study is the sidewall roughness measurement with a three-dimensional Atomic Force Microscopy (AFM) using tilted Z scanner. This 3D AFM is based on a decoupled XY and Z scanning configuration, in which the Z scanner can be intentionally tilted to the side. A sharp conical tip is typically used for imaging, which provides high resolution capability on both the flat surfaces (top and bottom) and the steep sidewalls. Keywords: AFM, critical dimension, LER, sidewall roughness, FinFET, EUV, Photoresist   1. INTRODUCTION The AFM has been widely used in universities for research purposes, and it has also been used in the industry as a metrology tool. One of the main applications of the AFM in metrology is height measurement. There are many different AFM height measurements; it could be a simple step height measurement, CMP dishing measurement, pole tip recession (PTR) measurement for hard disk read/write head [1], trench depth measurement, or some other type of measurement. The AFM is also used for surface roughness measurement. However, because most AFMs operate in a top-down configuration, the AFM has a limited access to the sidewall. This is especially true when the sidewall angle is close to, or greater than 90 degrees. To overcome this obstacle, certain AFM technology is able to use the side of a flared tip to image the sidewall [2]. However, the flared tip is quite blunt compared to the sharp conical tip, and it weakens the resolution on the sidewall. In addition, a flared tip is much more difficult to make, and its quality is hard to control.  A temporary solution has been used to characterize the sidewall roughness by mechanically cleaving the sample and then imaging the sidewall with AFM on tilted sample with regular conical tip. But this method only works for large features and it is destructive to the tip itself, as its sharpness fades in the process. Recently, Park Systems has introduced a new Park 3D AFM imaging technology using tilted Z scanner with a sharp conical tip for imaging [3]. When the tip tilts, it can easily reach the sidewall of the sample features,even when the sidewall angle is close to, or over 90 degrees. This new technology is extremely powerful for sidewall roughness measurement. In this paper, we have also studied the throughput and repeatability of this new Park 3D AFM technology as a metrology solution for sidewall roughness.    2.  Park 3D AFM WITH TILTED Z SCANNER The new Park 3D AFM is based on a decoupled XY and Z scanning system [4]. In this decoupled scanning configuration, the XY scanner is a 2D flexure scanner.  It moves the sample only in the XY direction and is independent from the Z scanner. The Z scanner is driven by a high force multi stack piezo element that moves the tip only in the Z direction (Figure 1-A).  The independent Z scanner allows the tip to be intentionally tilted to easily access the sidewall (Figure 1-B). The tilted scanner design allows CD measurement at the top, middle, and bottom of the lines, as well as roughness measurement along the line’s sidewall. The method builds upon the standard AFM tip design resulting in a technique that a) maintains the same resolution as the traditional AFM, b) that can be used with sharpened conical tips for best image resolution, and c) that does not suffer from corner inaccessibility from limited shape of tips with large radius. Figure 1, A. decoupled XY and Z scanning system; B, 3D AFM using tilting Z scanner     Figure 2.Imaging with Park 3D AFM   In order to take a full 3D image, the AFM typically scans at three tilted positions (center, left and right) to acquire three images. Each of these three images provides different information. For example, the image acquired at left tilted position provides detailed information on the left sidewall, as well as the top and bottom surface. These three images can be combined together to form a full 3D image of the feature (see Figure 2). The full 3D image provides detailed 3D information of the feature that is imaged, such as the top CD, middle CD, bottom CD, LER, LWR, sidewall angle, etc. [5]. Because a sharp conical tip is used, the tip can access the very small structures at the bottom of the feature, such as the small undercut at the very bottom of the photoresist line as seen in Figure 2.    3. SIDEWALL ROUGHNESS MEASUREMENT As Park 3D AFM uses the same feedback mechanism as regular AFM, it can also use the regular sharp conical tip for AFM imaging. One main factor which determines the imaging resolution is the sharpness of the tip. Compared with other types of tips (e.g. cylindrical tip or flared tip), the conical tip is best in sharpness that enables Park 3D AFM to provide the best resolution for the sidewall imaging. For sidewall imaging, the Z scanner can be fixed at one tilted position, and scan along the sidewall for significantly higher throughput. Figure 3 shows a Park 3D AFM image of a photoresist line. This image was taken by scanning in the direction along the line. The scan resolution is 256 pixels in X direction and 512 pixels in Y direction. This image shows high resolution information on both flat surfaces (top and bottom) and the sidewall. On the sidewall, the very rough and detailed grainy surface of the photoresist is visible, but the top and bottom surfaces are much smoother. Figure 3. 3D AFM image of the sidewall of a photoresist line   Single line profiles on the top, sidewall, and bottom surface can be extracted from the high resolution 3D AFM image. Figure 4 shows the single line profiles at different surface of the photoresist. These line profiles also clearly show that the sidewall is much rougher than the top and bottom surfaces.  Figure 4. Profiles of a photoresist line at different surfaces This Park 3D AFM sidewall image of the photoresist provides more detailed information about the sidewall surface than the traditional metrics (e.g. LER). This information is very useful in photoresist research and photolithography process development. Park 3D AFM with sharp conical tip not only provides high resolution image of the sidewall, it also measures the sidewall roughness with very high precision. Repeatability was tested with five sites and ten repeats. Table 1 shows the testing results. The average sidewall RMS roughness of the tested photoresist sidewall is about 6.1 nm. The Standard Deviation (STD) of the sidewall roughness measurement is about 0.20 nm or better for each individual site. And for the 5-site wafer average sidewall roughness, the measurement STD is 0.08 nm, which is only about 1.3% of the measured roughness.   Table 1. 5-site 10-repeat data of sidewall roughness measurement Figure 5 plots the 5-site average roughness of all the ten repeats. The plot shows that the roughness has stayed very consistent over the ten repeats, and there has been no pattern that shows it is increasing or decreasing during the repeated measurements. This means that the tip has remained as sharp as it was before the 50 measurements were done. Meanwhile, the soft photoresist was damaged only after ten repeated measurements. This is expected during the measurement because the AFM is operated in non-contact mode.   Figure 5.5-site average roughness of 10 repeats   During the repeatability test, each image was scanned for 32 lines along the sidewall direction, and each line has 512 data points. The image scan range is 0.5 um across the photoresist line, and 3um along the photoresist line. Each measurement took about two minutes. Of these two minutes, the image scanning took about one minute for each image (32 lines at 0.5 Hz). The sample navigation and alignment took another one minute, which included the stage moving, optical pattern recognition, AFM alignment, and tip approaching. This throughput could be further improved. When we scanned 32 lines over 0.5um across the photoresist line, about half of the line was on the top or bottom surface. Sixteen lines over 0.3um should have been sufficient to cover the sidewall surface with a decrease in the line density on the sidewall. By reducing the scan range along the photoresist line from 3.0um to 1.5um, the tip could scan at 1.0 Hz without compromising the surface tracking at the same settings. Hence, the scanning time for each image could be reduced to 16 seconds. With further optimization of the materials and procedure, the sample navigation and alignment time could be further reduced to about 45 seconds, resulting the throughput to  about 1minute/site.      4. CONCLUSIONS Park 3D AFM using tilted Z scanner with sharp conical tip provides high resolution sidewall imaging. It is also non-destructive and provides high throughput. Park 3D AFM can precisely measure the sidewall roughness with measurement STD of 0.08nm (5 site average) for the sidewall roughness of about 6.1 nm. Park 3D AFM can be used in applications, such as Fin-FET Device Characterization, EUV Photoresist Evaluation, Etching Process Fine Tuning, Photolithography Process Development, OPC Optimization, and OCD Calibration, among others.     ACKNOWLEDGEMENTS Thanks to Dr. Jane Frommer and Ms. Linda Sundberg at IBM Almaden Research Center for providing some interesting samples.     REFERENCES [1]        Joonhyung Kwon, Yong-Seog Kim, Kwanseok Yoon,Sang-Min Lee, Sang-il Park, "Advanced nanoscale metrology of pole-tip recession with AFM", Ultramicroscopy 105, 51–56 (2005) [2]        Bharat Bhushan, Harald Fuchs, and Masahiko Tomitori, Applied Scanning Probe Methods VIII, Springer, 31-75 (2008) [3]        Sang-Joon Cho, Jung-Min Lee, Byung-Woon Ahn, Joonhui Kim, Young Yoo, Yueming Hua, Sang-il Park, “Three-dimensional imaging of undercut and sidewall structures by atomic force microscopy”, Review of Scientific Instruments, 82, 023707 (2011) [4]        Joonhyung Kwon, Jaewan Hong, Yong-Seok Kim, Dong-Youn Lee, Kyumin Lee, Sang-min Lee, Sang-il Park, “Atomic force microscope with improved scan accuracy, scan speed,and optical vision”, Review of Scientific Instruments 74, 4378-4383 (2003) [5]                 Yueming Hua, Cynthia Coggins, Yong-ha Lee, Jung-min Lee, Kyung-deuk Ryang, Sang-il Park. “New 3-Dimensional AFM for CD Measurement and Sidewall Characterization”, Proceedings of SPIE (2011), Vol. 7971, 797118

24 Dec 2013

New 3-Dimensional AFM for CD Measurement and Sidewall Characterization

Yueming Hua, Cynthia Buenviaje-Coggins Park Systems Inc. 3040 Olcott St. Santa Clara, CA 95054, USA Yong-ha Lee, Jung-min Lee, Kyung-deuk Ryang, Sang-il Park Park Systems Corp. KANC 4F, Iui-Dong, 906-10 Suwon 443-270, Korea   ABSTRACT As the feature size in the lithography process continuously shrinks, accurate critical dimension (CD) measurement becomes more important. A new 3-dimensional (3D) metrology atomic force microscope (AFM) has been designed on a decoupled XY and Z scanner platform for CD and sidewall characterization. In this decoupled scanner configuration, the sample XY scanner moves the sample and is independent from the Z scanner which only moves the tip. The independent Z scanner allows the tip to be intentionally tilted to easily access the sidewall. This technique has been used to measure photoresist line patterns. The tilted scanner design allows CD measurement at the top, middle, and bottom of lines as well as roughness measurement along the sidewall. The method builds upon the standard AFM tip design resulting in a technique that a) maintains the same resolution as traditional AFM, b) can be used with sharpened tips for increased image resolution, and c) does not suffer from corner inaccessibility from large radius of curvature tips. Keywords: AFM, critical dimension, LER, LWR, sidewall, roughness   1. Introduction The atomic force microscope (AFM) is a high resolution imaging tool. AFM has been widely used in universities for research, and it has also been used in industry as a metrology tool. One of the main applications of AFM for metrology is height measurement. There are many different AFM height measurements. It could be a simple step height measurement, CMP dishing measurement, pole tip recession (PTR) measurement for hard disk read/write head [1], trench depth measurement, etc. AFM is also used for roughness measurements. Recently AFM has been used for angle measurement. Previously FIB-SEM was the main tool for angle measurement. With FIB-SEM, the sample is first cut with an ion beam (FIB), and then an image is taken of the cross section for the angle measurement with an electron beam (SEM). With AFM, the angle can be measured non-destructively with the angle determined from a series of scan lines. This creates an angle measurement technique which is faster, more accurate, and less costly than traditional angle measurements taken with FIB-SEM. AFM has also been used for CD measurement and sidewall imaging. This is one of the most challenging applications for AFM. As the feature size in the lithography process continuously shrinks, accurate CD measurement becomes more important. Just as nano-roughness became significant with ultrathin films, the sidewall roughness becomes significant in determining the functionality of devices with extremely small features. Accordingly the sidewall increasingly influences the CD measurement and characterizing the CD of a structure becomes more critical on the nanoscale. Over the past few years, AFM has become a powerful tool for accurate nanometrology. However, because most operate in a top-down configuration, AFM has limited access to the sidewall; this is especially true when the sidewall angle is near or greater than 90 degrees. To overcome this obstacle, special “boot shaped” tips have been developed with sophisticated detection algorithms to image sidewalls laterally [2], but this method has limited resolution Metrology, Inspection, and Process Control for Microlithography XXV, edited by Christopher J. Raymond, Proc. of SPIE Vol. 7971, 797118 · ⓒ 2011 SPIE · CCC code: 0277-786X/11/$18 · doi: 10.1117/12.879545   due to the large radius of curvature of the tip and blind spots at feature corners where the tip does not have sufficient physical access. This usually results in a rounded profile for the sharp corners at the bottom of features. A new 3-dimensional (3D) metrology AFM has been designed on a decoupled XY and Z scanner platform for CD and sidewall characterization [3].   2. Decoupled XY and Z Scanning System The new 3D AFM is based on a decoupled XY and Z scanning system [4]. In this decoupled scanning configuration, the XY scanner is a 2D flexure scanner. It moves the sample only in the XY direction and is independent from the Z scanner. The Z scanner is driven by a high force multi stack piezo element that moves the tip only in the Z direction (Figure 1-A). The independent Z scanner allows the tip to be intentionally tilted to easily access the sidewall (Figure 1-B). The tilted scanner design allows CD measurement at the top, middle, and bottom of the lines as well as roughness measurement along the line’s sidewall. The method builds upon the standard AFM tip design resulting in a technique that a) maintains the same resolution as traditional AFM, b) can be used with sharpened tips for increased image resolution, and c) does not suffer from corner inaccessibility from limited shape of tips with large radius. Figure 1, A. decoupled XY and Z scanning system; B, 3D AFM using tilting Z scanner   3. CD, LER, and LWR Measurement Using 3D-AFM A 165 nm 1:1 line/space photoresist dense line pattern was imaged with 3D-AFM. The line height is about 330 nm and the aspect ratio is about 2:1. Figure 2-A shows the cross section SEM image of the pattern. It shows that the two lines at both ends of the pattern collasped and leaned onto the the lines next to them. Figure 2-B is the 3D-AFM image of the pattern. X direction is the AFM fast scan direction. It shows that 3D-AFM is able to image all the surfaces (top, bottom, and both sidewalls with undercut) of the pattern. The AFM image matches very well the SEM image. The AFM image also clearly shows the wavy structure of the sidewall that is created by the standing wave of the light used for the photoresist exposure. Even the cracks at the bottom of the collasped lines are very clear on the 3D-AFM image. Figure 2-C is a single line profile of the AFM image. The line profile matches with the SEM image very well and shows that the sidewall angle is about 96 degrees, which suggests the sidewall has about 6 degrees of undercut. Figure 2. A, cross section SEM image of the photoresist dense line pattern; B, 3D-AFM image of the pattern; C, AFM line profile of the pattern   Figure 3. A-F, sliced 3D-AFM data at different Z heights; G, top-down SEM of the photoresist pattern With the high resolution 3D-AFM data, we can slice the data at different Z heights to measure the contour of the pattern. This capability is unique to 3D-AFM. Figure 3 shows the contour of the pattern at different Z heights, from 50nm to 300nm. The colored regions of these sliced images represent the photoresist. The two wider lines at both ends are formed by the collasped lines. These sliced images clearly show the rough edges of the photoresist lines, which are very similar to what we can see from the top-down SEM image (Figure 3 G).    Figure 4. CD measurement at different Z heights Figure 5. LER & LWR measurement at different Z heights With the sliced 3D-AFM data, we can calculate the CD, line edge roughness (LER), line width roughness (LWR) of the lines at different Z heights. Figure 4 shows the measured line CD (average of the 5 photoresist lines in the middle). From the data, we can see the line width is narrower (110.2 nm) at the bottom, gets wider as the height goes up, reaches its maximum line width (154.9 nm) at about 250nm height, and then decreases.. This trend of the line width change is consistent with the profile observed from the cross section SEM of the photoresist pattern (Figure 2-A). Figure 5. shows that LER and LWR do not vary with height. The LER is between 6.2nm and 6.9 nm, and LWR is between 8.6nm and 9.9nm.   4. Sidewall Imaging Using 3D-AFM A 300 nm semi-dense photoresist was also imaged with 3D-AFM with the Y direction (along the photoresist lines) as the fast scan direction. The advantage of scanning along the photoresist lines is that we can use high pixel resolution along the line without compromising the imaging speed.  Figure 6-A shows a 3D-AFM image. Because this image was taken with the Z scanner tilted to the left, the tip can image the top, bottom, and left sidewall of the pattern, but the right sidewall can not be imaged. The roughness image (topography image flattened by 1st order line-by-line in Y direction) is overlaid on the topography image to show the detail of the photoresist sidewall. The image shows that all three lines have similar sidewall roughness, and that the sidewalls are much rougher than the top and bottom surfaces. Figure 6-B is a direct view of the sidewall. This image clearly shows the grainy structure of the sidewall surface. High resolution line profiles can be extracted from the 3D-AFM sidewall image. Figure 7 shows single line profiles at the different locations indicated in Figure 6-A. The roughness at the different surfaces (top, bottom, and side) can be measured with these line profiles. The measured roughness is 0.50 nm for the bottom surface, 0.92 nm for the top surface, and 5.38 nm for the sidewall surface of the photoresist line. The measured sidewall roughness is directly related to the LER of the line.   Figure 6. A, 3D-AFM image of a 300 nm semi-dense photoresist line pattern showing the detail of the sidewall; B, a direct view at the sidewall showing the grainy structure of the sidewall surface   Figure 7. AFM line profiles at different surfaces shown in Figure 6A as dotted lines: A, top surface of a photoresist line; B, sidewall of a photoresist line; C, bottom surface of the sample 5. Conclusions This new 3D AFM is capable of advanced 3D imaging of both isolated lines and dense line patterns. The tip can easily reach the undercut sidewall and sharp corner at the bottom. This 3D AFM can be used for CD, LER, LWR, and sidewall roughness measurement. 3D-AFM is an excellent complementary technology for CD-SEM and FIB-SEM.   Acknowledgement We would like to thank Dr. Jane Frommer and Ms. Linda Sundberg at IBM Almaden Research Center for providing samples and helpful discussion. References [1]  Joonhyung Kwon, Yong-Seog Kim, Kwanseok Yoon,Sang-Min Lee, Sang-il Park, Advanced nanoscale metrology of pole-tip recession with AFM", Ultramicroscopy 105, 51?56 (2005) [2] Bharat Bhushan, Harald Fuchs, and Masahiko Tomitori, Applied Scanning Probe Methods VIII,Springer, 31-75 (2008) [3] Sang-Joon Cho, Jung-Min Lee, Byung-Woon Ahn, Joonhui Kim, Young Yoo, Yueming Hua,Sang-il Park, “Three-dimensional imaging of undercut and sidewall structures by atomic force microscopy”, Review of Scientific Instruments, 82, 023707 (2011) [4] Joonhyung Kwon, Jaewan Hong, Yong-Seok Kim, Dong-Youn Lee, Kyumin Lee, Sang-min Lee,Sang-il Park, “Atomic force microscope with improved scan accuracy, scan speed,and opticalvision”, Review of Scientific Instruments 74, 4378-4383 (2003)