We have demonstrated the unique capabilities of spectral interferometry (SI) with vertical traveling scatterometry algorithms (VTS) to solve 3D NAND challenges by measuring complex layer thicknesses of the multideck 3D structures directly from the VTS signals, without modeling, with Cell Over Periphery (COP) underlayer filtering. Multiple examples are presented in the paper, including the measurement of the thin and thick layers of memory structures above the complex logic arrays and the remaining thickness of the fully processed Si wafer from the back side after thinning. In addition, VTS and AI enable direct profiling of the deep through-type cell metal contacts in the areas with nonperiodic staircases and significant lateral variations under the measurement spot.
KEYWORDS: Semiconductor manufacturing, Inspection, Image classification, Scanning electron microscopy, Deep learning, Semiconductors, Classification systems
The semiconductor manufacturing process is becoming more complex and time-consuming due to smaller design rules and denser patterns, which inevitably leads to an increase in the number and types of defects. In the past, considerable efforts have been made to classify defects, and this has been implemented at the equipment level. However, in order to enhance the efficiency and productivity of semiconductor process development by automatically classifying random, systematic, and parametric defects according to various process schemes and structures, there is a need for a deep learning-based automatic defect classification technique with a higher degree of freedom and utilization. In this study, we used not only scanning electron microscope images, which have been actively studied, but also optical inspection images at various scales. Deep learning algorithms were evaluated for various layers of memory devices to select the optimal algorithm for each layer, and an accuracy of 94% or more was achieved, even with a small sample size (under 1000), which is critical in the R&D stage. It is expected that this technique will be able to spread and be applied to more diverse layers in the future. By providing faster and more diverse classifications of defects in semiconductor manufacturing processes and ensuring higher consistency through continuous sample size expansion, it is anticipated that this technique will contribute to shortening the development period and improving yield.
Historically when we used to manufacture semiconductor devices for 45 nm or above design rules, IC manufacturing yield was mainly determined by global random variations and therefore the chip manufacturers / manufacturing team were mainly responsible for yield improvement. With the introduction of sub-45 nm semiconductor technologies, yield started to be dominated by systematic variations, primarily centered on resolution problems, copper/low-k interconnects and CMP. These local systematic variations, which have become decisively greater than global random variations, are design-dependent [1, 2] and therefore designers now share the responsibility of increasing yield with manufacturers / manufacturing teams. A widening manufacturing gap has led to a dramatic increase in design rules that are either too restrictive or do not guarantee a litho/etch hotspot-free design. The semiconductor industry is currently limited to 193 nm scanners and no relief is expected from the equipment side to prevent / eliminate these systematic hotspots. Hence we have seen a lot of design houses coming up with innovative design products to check hotspots based on model based lithography checks to validate design manufacturability, which will also account for complex two-dimensional effects that stem from aggressive scaling of 193 nm lithography. Most of these hotspots (a.k.a., weak points) are especially seen on Back End of the Line (BEOL) process levels like Mx ADI, Mx Etch and Mx CMP. Inspecting some of these BEOL levels can be extremely challenging as there are lots of wafer noises or nuisances that can hinder an inspector’s ability to detect and monitor the defects or weak points of interest. In this work we have attempted to accurately inspect the weak points using a novel broadband plasma optical inspection approach that enhances defect signal from patterns of interest (POI) and precisely suppresses surrounding wafer noises. This new approach is a paradigm shift in wafer inspection by leveraging systematic defect locations for high sensitivity inspection, thereby enhancing the discovery and monitoring of yield-limiting defects at traditional optical inspection throughput.
This paper presents a methodology for detecting defects more effectively that have a substantial yield impact on several critical layers using a simulation program, which is considerably helpful in analyzing defects on the wafer. First, this paper presents a simple analysis method that uses mathematical treatment for multi thin film layers. This instantly gives us a highly intuitive idea for selecting an inspection mode based on the reflectivity and transmittivity. Second, we introduce numerical method for wafer defect of interest with finite difference time domain (FDTD) method, and provide correlation between the expectation and experimental results. The goal of these studies is to determine the feasibility of implementing theoretical approaches with numerical method at wafer defect inspection. Overall, this paper discusses the effective wafer inspection methodology and the advantages of defect simulation with numerical analysis at semiconductor manufacturing for accelerated development of advanced design node devices.
Development of an advanced design node for NAND flash memory devices in semiconductor manufacturing requires
accelerated identification and characterization of yield-limiting defect types at critical front-end of line (FEOL) process
steps. This enables a shorter development cycle time and a faster production ramp to meet market demand. This paper
presents a methodology for detecting defects that have a substantial yield impact on a FEOL after-develop inspection
(ADI) layer using an advanced broadband optical wafer defect inspector and a scanning electron microscope (SEM)
review tool. In addition, this paper presents experimental data that demonstrates defect migration from an ADI layer to
an after-clean inspection (ACI) layer, and provides clear differentiation between yield-impacting critical defects and noncritical
defects on the layers. The goal of these studies is to determine the feasibility of implementing an inspection point
at ADI. The advantage of capturing yield-limiting defects on an ADI layer is that wafers can be reworked when an
excursion occurs, an option that is not always possible for ACI layers. Our investigation is divided into two parts: (1)
Inspection of an ADI layer with high sensitivity to find an accurate representation of the defect population and to gain
understanding on the propagation of defects from the ADI layer to the ACI layer; and, (2) Inspection of an ACI layer to
develop an understanding of unique defects generated by the ACI process step. Overall, this paper discusses the
advantages of baselining defectivity at ADI process levels for accelerated development of advanced design node memory
devices.
Distributions of via Depth and bottom CD for TSV wafer have been studied by scanning interference microscopy
(Unifire 7900, Nanometrics Inc.). We plotted whole wafer maps for each via depth and bottom CD and found useful
relationship between them (i.e. via depth is in inverse proportion to bottom CD in general). Average values of via depth
and bottom CD are ~60um and ~4um and their standard deviation values are 1.28% and 5.14% respectively. We also
demonstrated kinds of defects at via top (or bottom) which can cause disturbance of total via count during 3D inspection.
Our results can be a good introduction to scanning interference tool as a monitoring tool for TSV high volume
manufacturing.
Continuing demand for high performance microelectronic products propelled integrated circuit technology into 45 nm
node and beyond. The shrinking device feature geometry created unprecedented challenges for dimension metrology in
semiconductor manufacturing and research and development. Automated atomic force microscope (AFM) has been used
to meet the challenge and characterize narrower lines, trenches and holes at 45nm technology node and beyond. AFM is
indispensable metrology techniques capable of non-destructive full three-dimensional imaging, surface morphology
characterization and accurate critical dimension (CD) measurements. While all available dimensional metrology
techniques approach their limits, AFM continues to provide reliable information for development and control of
processes in memory, logic, photomask, image sensor and data storage manufacturing. In this paper we review up-todate
applications of automated AFM in every mentioned above semiconductor industry sector. To demonstrate benefits
of AFM at 45 nm node and beyond we compare capability of automated AFM with established in-line and off-line
metrologies like critical dimension scanning electron microscopy (CDSEM), optical scatterometry (OCD) and
transmission electronic microscopy (TEM).
The amplitude setpoint affects the critical dimension measurement with CD-AFM. The setpoint amplitude is the
amplitude of the resonant oscillation of the AFM tip maintained by the feedback loop as it scans the surface. The
Setpoint therefore decides the tip-surface distance, and the tip-surface interaction force as well. Normally, the tip moves
an unknown distance away from the sample surface. Such a tip-sample distance on the top and bottom surface is
cancelled out in height measurement. In width measurement, however, the tip-sample distance on the left and right
sidewall will add up to produce a bias in the measured CD values. The bias will appear in the opposite way and by the
same amount in line and trench measurement. We conducted the experiments to see the effect, and found out there
exists the dependence of the measured linewidths on the setpoint in the consistent behavior as our hand-waving predicts.
The effect may be a significant uncertainty source in the CD-AFM metrology.
The measurement of edge roughness has become a hot issue in the semiconductor industry. Especially the contact roughness is being more critical as design rule shrinks. Major vendors offer a variety of features to measure the edge roughness in their CD-SEMs. For the line and space patterns, features such as Line Edge Roughness (LER) and Line Width Roughness (LWR) are available in current CD-SEMs. However the features currently available in commercial CD-SEM cannot provide a proper solution in monitoring the contact roughness. We had introduced a new parameter R, measurement algorithm and definition of contact edge roughness to quantify CER and CSR in previous paper. The parameter, R could provide an alternative solution to monitor contact or island pattern roughness. In this paper, we investigated to assess optimum number of CD measurement (1-D) and fitting method for CER or CSR. The study was based on a circular contact shape. Some new ideas to quantify CER or CSR were also suggested with preliminary experimental results.
The measurement of edge roughness has become a hot issue in the semiconductor industry. Major vendors offer a variety of features to measure the edge roughness in their CD-SEMs. However, most of the features are limited by the applicable pattern types. For the line and space patterns, features such as Line Edge Roughness (LER) and Line Width Roughness (LWR) are available in current CD-SEMs. The edge roughness is more critical in contact process. However the measurement of contact edge roughness (CER) or contact space roughness (CSR) is more complicated than that of LER or LWR. So far, no formal standard measurement algorithm or definition of contact roughness measurement exists. In this article, currently available features are investigated to assess their representability for CER or CSR. Some new ideas to quantify CER and CSR were also suggested with preliminary experimental results.
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.