Most advanced wafer fabs have embraced complex pattern decoration, which creates numerous challenges during in-fab reticle qualification. These optical proximity correction (OPC) techniques create assist features that tend to be very close in size and shape to the main patterns as seen in Figure 1. A small defect on an assist feature will most likely have little or no impact on the fidelity of the wafer image, whereas the same defect on a main feature could significantly decrease device functionality. In order to properly disposition these defects, reticle inspection technicians need an efficient method that automatically separates main from assist features and predicts the resulting defect impact on the wafer image. Analysis System (ADAS) defect simulation system[1]. Up until now, using ADAS simulation was limited to engineers due to the complexity of the settings that need to be manually entered in order to create an accurate result. A single error in entering one of these values can cause erroneous results, therefore full automation is necessary. In this study, we propose a new method where all needed simulation parameters are automatically loaded into ADAS. This is accomplished in two parts. First we have created a scanner parameter database that is automatically identified from mask product and level names. Second, we automatically determine the appropriate simulation printability threshold by using a new reference image (provided by the inspection tool) that contains a known measured value of the reticle critical dimension (CD). This new method automatically loads the correct scanner conditions, sets the appropriate simulation threshold, and automatically measures the percentage of CD change caused by the defect. This streamlines qualification and reduces the number of reticles being put on hold, waiting for engineer review. We also present data showing the consistency and reliability of the new method, along with the impact on the efficiency of in-fab reticle qualification.
Even small defects on the main patterns can create killer defects on the wafer, whereas the same defect on or near the decorative patterns may be completely benign to the wafer functionality. This ambiguity often causes operators and engineers to put a mask "on hold" to be analyzed by an AIMS™ tool which slows the manufacturing time and increases mask cost. In order to streamline the process, mask shops need a reliable way to quickly identify the wafer impact of defects during mask inspection review reducing the number of defects requiring AIMS™ analysis.
Source Mask Optimization (SMO) techniques are now common on sub 20nm node critical reticle patterns These techniques create complex reticle patterns which often makes it difficult for inspection tool operators to identify the desired wafer pattern from the surrounding nonprinting patterns in advanced masks such as SMO, Inverse Lithography Technology (ILT), Negative Tone Development (NTD).
In this study, we have tested a system that generates aerial simulation images directly from the inspection tool images. The resulting defect dispositions from a program defect test mask along with numerous production mask defects have been compared to the dispositions attained from AIMS™ analysis. The results of our comparisons are presented, as well as the impact to mask shop productivity.
IC fabs inspect critical masks on a regular basis to ensure high wafer yields. These requalification inspections are costly
for many reasons including the capital equipment, system maintenance, and labor costs. In addition, masks typically
remain in the “requal” phase for extended, non-productive periods of time. The overall “requal” cycle time in which
reticles remain non-productive is challenging to control. Shipping schedules can slip when wafer lots are put on hold
until the master critical layer reticle is returned to production. Unfortunately, substituting backup critical layer reticles
can significantly reduce an otherwise tightly controlled process window adversely affecting wafer yields.
One major requal cycle time component is the disposition process of mask inspections containing hundreds of defects.
Not only is precious non-productive time extended by reviewing hundreds of potentially yield-limiting detections, each
additional classification increases the risk of manual review techniques accidentally passing real yield limiting defects.
Even assuming all defects of interest are flagged by operators, how can any person's judgment be confident regarding
lithographic impact of such defects? The time reticles spend away from scanners combined with potential yield loss due
to lithographic uncertainty presents significant cycle time loss and increased production costs
An automatic defect analysis system (ADAS), which has been in fab production for numerous years, has been improved
to handle the new challenges of 14nm node automate reticle defect classification by simulating each defect’s printability
under the intended illumination conditions. In this study, we have created programmed defects on a production 14nm
node critical-layer reticle. These defects have been analyzed with lithographic simulation software and compared to the
results of both AIMS optical simulation and to actual wafer prints.
Advanced IC fabs must inspect critical reticles on a frequent basis to ensure high wafer yields. These necessary requalification inspections have traditionally carried high risk and expense. Manually reviewing sometimes hundreds of potentially yield-limiting detections is a very high-risk activity due to the likelihood of human error; the worst of which is the accidental passing of a real, yield-limiting defect. Painfully high cost is incurred as a result, but high cost is also realized on a daily basis while reticles are being manually classified on inspection tools since these tools often remain in a non-productive state during classification. An automatic defect analysis system (ADAS) has been implemented at a 20nm node wafer fab to automate reticle defect classification by simulating each defect’s printability under the intended illumination conditions. In this paper, we have studied and present results showing the positive impact that an automated reticle defect classification system has on the reticle requalification process; specifically to defect classification speed and accuracy. To verify accuracy, detected defects of interest were analyzed with lithographic simulation software and compared to the results of both AIMS™ optical simulation and to actual wafer prints.
IC fabs inspect critical masks on a regular basis to ensure high wafer yields. These requalification
inspections are costly for many reasons including the capital equipment, system maintenance, and
labor costs. In addition, masks typically remain in the “requal” phase for extended, non-productive
periods of time. The overall “requal” cycle time in which reticles remain non-productive is
challenging to control. Shipping schedules can slip when wafer lots are put on hold until the master
critical layer reticle is returned to production. Unfortunately, substituting backup critical layer
reticles can significantly reduce an otherwise tightly controlled process window adversely affecting
wafer yields.
One major requal cycle time component is the disposition process of mask inspections containing
hundreds of defects. Not only is precious non-productive time extended by reviewing hundreds of
potentially yield-limiting detections, each additional classification increases the risk of manual
review techniques accidentally passing real yield limiting defects. Even assuming all defects of
interest are flagged by operators, how can any person's judgment be confident regarding lithographic
impact of such defects? The time reticles spend away from scanners combined with potential yield
loss due to lithographic uncertainty presents significant cycle time loss and increased production
costs.
Fortunately, a software program has been developed which automates defect classification with
simulated printability measurement greatly reducing requal cycle time and improving overall
disposition accuracy. This product, called ADAS (Auto Defect Analysis System), has been tested in
both engineering and high-volume production environments with very successful results. In this
paper, data is presented supporting significant reduction for costly wafer print checks, improved
inspection area productivity, and minimized risk of misclassified yield limiting defects.
EUV lithography is expected to begin production in 2014. Production of successful EUV photomasks requires patterned
mask inspection (PMI). The ultimate PMI tool is expected to utilize actinic (EUV) illumination. Development of such a
tool is expected to require three years after funding. Current test EUV masks, such as 22 nm, can be inspected using 193
nm wavelength deep UV (DUV) inspection tools similar to those currently being used for DUV masks. The DUV
inspection tools may be extended for the 16 nm node. However EUV production is expected to start with 11 nm node
masks which cannot be inspected with proposed DUV inspection tools. Therefore E-beam inspection (EBI) is discussed
as the interim PMI method.
EBI has the advantage of high resolution and the disadvantages of low inspection speed and relative insensitivity to ML
defects (in the multi-layer material). EBI inspection speed is limited by the pixel size, pixel capture rate and the number
of electron columns. The pixel rate is limited by the detector time-resolution, the beam current, and the detection
efficiency.
Technical improvements in beam focus, secondary electron detection, and defect detection and analysis provide good
performance for 22 nm node masks. We discuss the advances and show that performance can be extrapolated for 16 and
11 nm node patterned mask inspections.
We present sensitivity and false-defect frequency results of using the Holon EBI tool on 22 nm test masks and a roadmap
for extending its operation for use on 16 and 11 nm node masks for inspections requiring 2-5 hours per mask.
Mask defect disposition gets more difficult and time-consuming with each progressive lithography node. Mask
inspection tools commonly use 250 nm wavelength, giving resolution of 180 nm, so critical defect sizes are far less than
the optical resolution - too small for defect analysis. Thus the rate of false or nuisance defect detection is increasing
rapidly and analysis of detected defects is increasingly difficult. As to judging the wafer printability of defects, AIMS
(Aerial Image Measurement System) tools are commonly used but are also time-consuming if defect count is high. For
improving the efficiency of mask defect disposition, we propose the combination of a SEM defect review tool and defect
disposition and simulation software, which use high-resolution SEM images of defects to do defect review, defect
disposition, and wafer printing simulation of defects automatically or manually.
The SEM defect review tool, DIS-05 developed by Holon Co. Ltd., is designed for defect review and disposition using
reference images derived from e-beam files or CAD database. This tool uses the Automated Defect Analysis Software
(ADAS) developed from AVI LLC. to interface the inspection tool and the DIS-05. ADAS detects false defects before
SEM imaging and performs aerial image simulation from the SEM and CAD images to estimate the wafer CD error
caused by each defect. We report on its speed (>300 defects/hour), classification accuracy and simulation accuracy when
used with masks at the 45 nm technology node and beyond. This combination of SEM and ADAS is expected to
significantly accelerate process development and production for the 45 and 32 nm nodes. It will also increase the masksper-
day throughput of inspection and AIMS tools by shifting most defect review to ADAS software using SEM images.
At preliminary tests showed the combination tool can do auto defect disposition and simulation with promising results.
ADAS (Automated Defect Analysis Software) is the first product to fully automate mask defect analysis for mask shops
and fabs. ADAS classifies and dispositions photomask defects quickly and accurately. Disposition is based on defect
size and printability measurements from simulation.
Full analysis of inspection reports with 100 defects requires 2 seconds. Printability measurements match AIMS within 6
percent at 3 sigma on 45 nm test masks. Repeatability is 5 percent at 3 sigma over multiple inspections. ADAS can
reduce the need for production AIMS measurements by 90% and eliminate operator review errors and the repelliclizations
they cause. ADAS increases overall inspection efficiency for mask shop first-inspection and final
inspection. It can automate fab requalification inspections and eliminate the need for incoming inspection.
Reducing defect repairs and accelerating defect analysis is becoming more important as the total cost of defect repairs
on advanced masks increases. Photomask defect specs based on printability, as measured on AIMS microscopes has
been used for years, but the fundamental defect spec is still the defect size, as measured on the photomask, requiring the
repair of many unprintable defects.
ADAS, the Automated Defect Analysis System from AVI is now available in most advanced mask shops. It makes the
use of pure printability specs, or "Optimal Defect Specs" practical. This software uses advanced algorithms to eliminate
false defects caused by approximations in the inspection algorithm, classify each defect, simulate each defect and
disposition each defect based on its printability and location.
This paper defines "optimal defect specs", explains why they are now practical and economic, gives a method of
determining them and provides accuracy data.
The detection, classification and disposition of defects is an important function that commands significant resources in mask making. Current processes use manual evaluation of defects, which is slow, subject to errors, and provides sparse data for process improvement. The automated defect analysis software described here reads inspection reports from mask inspection tools, classifies each defect, and measures both its size and printability. It combines and compares data from multiple inspections to provide critical process development data. Data from 144 masks is presented showing that the system missed no critical defects found by operators. These inspections also demonstrated numerous occasions for improved classifications compared to that given by the operators. This capability gives improved disposition, an easy path to using simulator based printability for disposition, and significant improvements in mask yield.
Reticle pinhole defects below 200nm are problematic from several standpoints. The wafer manufacturer presents a specification to the reticle producer, who in turn charges the inspection tool vendors with the task of detecting pinholes of a given size. The measurement of these pinholes, especially on programmed defect test masks, becomes critical to the success of this flow. Measuring the size of these small pinholes using the current SEM method often produces inconsistent results when compared to pinhole printability. Early studies have suggested that since the SEM measures only the top surface of the pinhole, the measurement does not account for edge wall angle and partial filling which reduces the pinhole transmission and subsequent printability. This investigation focuses on several transmitted light approaches for reticle pinhole measurement on programmed defect masks. An attempt to correlate these methods back to traditional SEM and optical sizing methods will also be attempted.
Lithographers' ability to set useful defect and contact specifications has almost disappeared as chip geometries have shrunk. As features sizes have decreased, measurement error has increased to 25% of the maximum allowable defect size. This has made defect disposition so difficult that many processes now require that all detected defects be repaired because the automatic defect sizing is almost meaningless, that is, the required guard band is nearly the size of the defect specification (Reynolds, BACUS 2000). Many mask processes have abandoned defect sizingin favor of stepper simulation, either using simulation microscope, such as AIMS, or software, such as NTI's VSS. However, AVI's optical Flux-Area measurement technique provides accuracy and repeatability that gives the simple, time tested defect specification technique new life. This study demonstrates high quality edge-, contact-, and OPC- defect disposition can be achieved using the Flux-Area technique. A test mask with a range of edge defects as well as mis-sized contacts and OPC defects was written. The mask defect sizing performed with the AVI is shown to be consistent on different chips using the same process. Thus it is shown that all the over-spec defects on the wafer were measured as over-spec on the mask. Results show that edge defect size on the wafer can be accurately predicted from the AVI defect area; that printed contact size is linearly proportional to the AVI measured area, on both square and irregular contacts; and that OPC defects (printed line-end separation errors) can be accurately predicted from AVI serif-area measurements on the mask. With the Flux-Area measurement technique as implemented on the AVI Photomask Metrology System, defects can be measured with long term repeatability and rms repeatability between machines of better than 10nm, 3% of a 0.3micrometers defect. This means that guard bands can often be reduced from 0.15micrometers to below 0.05micrometers .
Metrology is essential to success in all manufacturing processes. In microlithography metrology techniques have begun to shift from optical to SEM. In this paper we compare the capabilities of the new Flux-Area optical technique and of SEM techniques. SEM measurement has been increasing in mask shops because of the higher resolution it provides, despite disadvantages including damage to masks, charging effects, and inability to operate with pellicles. Optical measurements of photomasks are preferred because they are performed with light, in transmission. The success of virtual stepper software, which uses optical images in simulating stepper output, has demonstrated that optical images contain sufficient information to predict the results of even subwavelength lithography. Flux-Area measurement allows optical instruments to accurately measure features as small as (lambda) /6, or 0.08 (mu) using visible light (Fiekowsky and Selassie, 1999). It also allows the measurement of Optical Dimensions. This is the width of a line defined by the flux of light it transmits to the objective lens. In this study Flux-Area measurements of linewidths and contact are compared to SEM measurements and DUV AIMS microscope images. Results show that Flux-Area measurements are linear down to the smallest lines and holes tested, 0.23 (mu) and 0.4(mu) respectively. Thus the Flux-Area technique provides a practical alternative to SEM for metrology on current and future generation photomasks.
Small contact holes are the most difficult structures for microlithography to print because their sizes match the resolution of the stepper. Thus a mask hole that is a little too small won't print at all. It is widely accepted that the size of the contact hole on the wafer is controlled by the total light flux that gets through the contact hole on the mask. This is demonstrated by the fact that normal small contact holes print round on the wafer, no matter what shape they were on the mask. Therefore it is the area, not the shape that affects the wafer print. Contact holes have traditionally been measured in height and width. Recently, technology has been introduced to estimate corner rounding, so that a combination of height, width, and corner rounding could be used to estimate the area of the contact. Th introduction of the flux-area technique allows direct measurement of the optical area of features as small as (lambda) /4, e.g. 0.15(mu) using visible light. These measurements have been shown to be linear, repeatable, and highly correlated to what prints on the wafer. This study examines the accuracy and repeatability of optical contact hole area measurements using visible light and I-line illumination with the AVI Metrology system. A test mask is designed and printed by Photronics. It is measured with the AVI Photomask Metrology system taking images from a KLA353uv inspection system and a Zeiss AIMS tool, and a SEM. The mask is then printed at LSI Logic on a 0.25micrometers process. The contact sizes on the wafer are measured on a SEM. Results show that optical measurements from the AVI system correlate to the printed contact size as well as SEM measurements of the mask, and twice as well as conventional optical measurements. Furthermore, optical mask measurements at the mask shop and the fab correlate within 20nm despite the use of different wavelengths, and measurement with and without pellicles.
Soft defects on photomasks have, historically, been difficult to measure, and predict how the measured size of a soft defect will correlate to what prints, if at all. Over the past few years KLA-Tencor STARlight surface inspection has become the inspection of choice for soft defects. Though the capture rate of this tool is exceptional, the defect sizing capability has lacked in accuracy. Customer specifications have traditionally been built around defect size and transmission. If a given defect cannot be accurately sized then it cannot be accurately dispositioned. In this study we are attempting to show a correlation between the AVI defect measurement tool sizing and what actually prints on the wafer. We will show defect sizing both from the KLA-Tencor STARlight and pattern tools, the AVI tool, AIMS and VSS printability data.
The patented Flux-Area technique of metrology with optical images has been proven to provide accurately and repeatable measurements of defect sizes as small as 0.08 (mu) , and is in use in 12 leading edge mask shops and wafer fabs around the world. This paper describes the extension of this technique for linewidth measurement and the result of test on photomasks, using lines as narrow as 0.25 micrometers . Linewidths were measured with SEM and optical images analyzed with the Flux-Area technique. Results show that the new technique provides linear measurements on even the smallest lines, using visible as well as UV illumination. This technique promises to allow mask makers to continue measuring their masks optically, even as linewidths shrink much smaller than the optical wavelengths used in the measurement. Further, this technique allows older visible light system to be used for measuring even DUV masks. Finally, this technique does not require thresholds: it only requires an image and that the optical magnification be known. The technique simply measures the amount of light absorbed by a feature, the fundamental optical quality of any photomask feature.
The search for the 'holy grail' of a fast, reliable, inexpensive predictor of defect printability has reached a new level. Taking images from several inspection tools (KLA-351, KLA-353, and KLA Starlight) during defect review, the AVI Photomask Metrology System provides measurement repeatability better than 5 nm, and significantly better correlation to printability than Scanning Electron Microscope (SEM) measurements. SEM measurements of printed defects are compared to various measurements of the defects on the mask. Analysis shows, as expected, that optical area measurements provide the best correlation to printability. Further, images from existing inspection tools are shown to be sufficient to produce these measurements using AVI's new 'Flux-area' technique.
A highly repeatable and accurate mask defect measurement has been developed. Defects from 0.1 to 1.5 microns in diameter are measured by computing the total light absorbed or transmitted by a defect. RMS repeatability of better than 9 nanometers on 0.4 micron defects has been achieved. Reliable measurement of defect size is important for developing lithography technologies for smaller geometries, and for commercial mask production. As mask feature sizes have dropped below the wavelength of visible light, getting reliable defect size measurement has become nearly impossible. Even scanning electron microscope (SEM) and atomic force microscope (AFM) measurements have not yet proven reliable even though they provide resolutions down to a few nanometers. This technique of measuring flux absorption or transmission allows reliable measurement of defects that are several times smaller than the wavelength of light used to examine them, with repeatability of 2 - 10 nanometers, depending on the image source. Transmitted light images are acquired from KLA-3xx, Starlight, KLA-219, DRS-1, DRS-2, or other video microscopes. Then the amount of light flux absorbed (by a spot or chrome extension), or transmitted (by a hole or clear intrusion) is measured. That change in flux is converted to an area, which can then be converted to a diameter. This system is currently in use in several large mask shops. It promises to be a powerful QA and analysis tool for developing masks for .25 micron and smaller geometries. Accuracy and repeatability tests have been performed on reference defects on Dupont VeriMasks, and using PSL spheres. Repeatability is limited by vibration of the image and by pixel artifacts in the images from KLA-3xx machines. Accuracy cannot be objectively assessed because there is no 'NIST traceable' reference for defect sizes. However, chrome defect size appears to be linearly correlated to absorbed or transmitted flux, as one would expect from the physics, so defect area accuracy is expected to be similar to the repeatability, around 10 nanometers. This technique does not easily provide separate x- and y- dimensions for non-round defects smaller than the wavelength of light used in the microscope (typically 0.5 micron). Larger defects can be measured in two dimensions using conventional techniques, and Fourier transform techniques can be used to provide useful estimates of x- and y- dimensions of smaller defects.
An objective and practical method for accurate sizing of mask defects found with inspection tools is presented. Diameters of defects found on a KLA Instrument's 219e mask inspection tool were measured to an accuracy of 0.10 microns by using gray scale image information provided by the KLA machine. Final through-the-pellicle inspection may include defects which are smaller than specification. Precise and practical measurement of these defects is vital for objective disposition. Further, the ability to perform statistics on these measurements allows greater process improvement and inspection strategy tuning. Test defects were measured on a Verimask (a plate with programmed defects and 'standard' sizes for each defect, made by Dupont) and supplied as a calibration gauge for the measurement tool. Images of the defects were extracted from the video output of the KLA tool. A reference image was derived by averaging the light intensity in the region surrounding the defect. In cases where the surrounding area was non-uniform (edge defect), a separate reference image was extracted from the KLA tool. Then the area of the defect was measured by subtracting the reference image from the original and measuring the total difference in light energy in the region of the defect. This total light difference was scaled to the calibration measurements by using a quadratic least squares fit. Separate calibration curves are used for spots, holes, intrusions, and extensions, correcting for optical characteristics of the system and likely skews in the reference measurements.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.