Europe's SEA Advances Analytic and Metrology Tools
Jeff Bruchez, Semiconductor Equipment Assessment, Oxon, UK -- Semiconductor International, 3/1/2003
|
Rapidly shrinking device dimensions and concurrent increases in process complexity necessitate the application of more sophisticated yet cost-effective analytic/metrology techniques for process development, ramp up and finally for use in mainstream production lines. At the manufacturing level, the trend is toward equipping with appropriate characterization tools within each wafer stage for in situ/in-line/at-line non-destructive whole-wafer analysis inspection and defect review.
Although several analytic methods exist, performance needs to be progressively upgraded to satisfy the demands of the ITRS. Innovation is needed for in situ process characterization and effective defect review, including physical/chemical metrology characterization techniques capable of identifying the exact nature of impurities — size, shape, composition, location, and their subsequent correlation with device yield — to solve major issues for defect reduction. The size of killer defects is also rapidly shrinking, so microscopes, analytic SEM and microanalysis techniques must keep pace with this requirement.
Ideally, the equipment should be suited to in situ or in-line/at-line use rather than embedded in expensive, unwieldy "laboratory-bound" equipment. To do this, these complex tools must engender innovative concepts while being reliable, easy-to-use and cost-effective in terms of their cost of ownership (CoO), value of ownership (VoO) or return on investment (ROI).
The main goal of the Semiconductor Equipment Assessment (SEA) program and its projects is to address the innovation-equipment productivity gap by bringing together key IC manufacturers and equipment companies to work on productivity and user-friendliness while state-of-the-art applications are addressed. SEA covers all major semiconductor manufacturing issues, with more than 60 projects to date (www.sea.rl.ac.uk) and a significant number of these projects concern analysis/metrology. By example, this article focuses on recent results from four such projects. It also gives information on some of the new projects started in 2002 that will provide results during 2003.
APC for 300 mm plasma (APC300)The Hercules advanced process control (APC) plasma monitor tool from ASI (Berlin) has been successfully evaluated for in situ analysis on Infineon's SC300 line (Dresden, Germany), and by the project partners AMD Saxony (Dresden), STMicroelectronics (Rousset, France) and Infineon (Regensburg, Germany). It was installed on equipment from leading suppliers and synchronized with CIM and FDC systems, showing multiple technical and economic benefits. It uses self-excited electron resonance spectroscopy (SEERS) to yield independent plasma parameters: electron density, electron/neutrals, collision rate, bulk power, and rf peak voltage.
Engineering cooperation with equipment manufacturer Lam Research has enabled integration of Hercules with the Lam Domino plug-and-play interface to provide synchronized data flow between sensor and FDC, enabling rapid fault analysis. Equipment improvements that have been implemented include:
- Improved acquisition mode for Applied Materials' eMxP+ series.
- Communication and data interfaces that now accept data from Triant, SI Automation, Brookside and APC Trend.
- Improved software for faster data analysis, and measurement up to 10× faster.
| 1. A batch of 23 wafers during oxide
etch shows the high collision rate of a single low-yield
wafer. |
Figure 2 shows the collision rate for SiO2 test wafers for a critical process at SC300. There are short-term upward drifts for each wafer associated with a quartz ring temperature effect around the wafer, but there is a longer-term downward drift that was identified as improper heating of the chamber lid.
| 2. Electron collision rate and lid temperature of six SiO2 test wafers for a critical process at Infineon’s SC300. |
The <M5k> from MueTec GmbH (Munich) is an automated optical metrology system for measuring reticle and photomask critical dimensions (CDs). It has been evaluated at Compugraphics International (Glenrothes, UK), with partners including Infineon's mask house, Taiwan Mask Corp. (Hsinchu, Taiwan), and IMEC (Leuven, Belgium).
The system operates in a stepper-like mode with 248 nm light using a high-numerical-aperture (NA) Leica microscope objective featuring an optical resolution of better than 90 nm. Through-pellicle metrology was achieved with a long-working-distance objective lens.
| 3. CD measurement on 89 nm lines (300
nm pitch). |
An important aspect is that the measurement principle is very similar to the application of the photomask in the final exposure system, so interpretation of DUV optical microscope images is less ambiguous than with SEM or atomic force microscopy (AFM). The influence of variables such as edge slope, thin residuals of the masking layer or a phase shift are inherently taken into account. The system identified effects that were found to be undetectable by SEM assessment such as lack of edge acuity caused by wavy edges on lines. Significantly, through-pellicle measurement and printability simulation of mask features or defects are also only possible with the optical approach.
The <M5k> also proved to be unequalled in terms of its measurement repeatability for COG and phase-shift masks (PSMs) in a production environment. Two <M5k> systems were tested with both achieving long-term repeatability of 0.8 and 1.3 nm (3s), routinely meeting the target specification of 1.5 nm. This outperforms SEM and AFM methods.
Measurements on 193 nm MoSi EAPSMs (PSMs for the 80 nm node) gave extremely good results using the high-NA version of the <M5k>, a major breakthrough since this material offers insufficient contrast for current i-line metrology systems, and is very difficult to measure on CD-SEM.
When configured with a high-NA (150×/0.90 objective) to a long-working-distance version (LWD, 50×/0.55), it enables through-pellicle measurement with high image contrast. The decrease of practical resolution was shown to be within acceptable limits such that COG lines can be measured with a long-term repeatability of 1.0 nm (3s) down to 400 nm. The equivalent figure down to ~150 nm is still 3.0 nm (3s), which is an excellent result bearing in mind that this is the specified repeatability of current reticle CD-SEMs.
With respect to usability, the <M5k> system has proven to be a tool that can be readily integrated into the manufacturing environment for automatic operation. The software includes tools for remote job setup and execution and to measure lines/spaces, edge roughness, optical proximity corrections (OPCs), contacts, corner rounding and interface to CATS Mask Data Fracturing Software. It can also provide mask review and defect printability analysis after automatic defect inspection with associated analytic tools.
Remote setup of measurement jobs took 2-4 min for up to 100 measurement locations, leading to an overall throughput of more than three reticles/hr. With this performance, a single <M5k> could comfortably support the CD metrology requirements for 10 pattern generators. Initial tests with a newly developed interface to CATS software have shown that the time necessary for job preparation can be decreased to about 15 sec.
The reliability of the tool has been very good over months, reaching excellent figures during the final test period of one month with an uptime of 98.9%, MTBF of 1200 hr, and MTTR of 1.0 hr. Also, excellent long-term measurement stability was achieved with a 3s value of 0.5 nm over a total of 140 measurements distributed over 14 days.
The McDOR project has met all its objectives and the overall experience has greatly proven the suitability of the <M5k> for 90 nm requirements and its potential for the 65 nm node. Its excellent performance is supported by a CoO at roughly half that of a current CD-SEM.
Microcalorimeter EDS assessment (MESA)A quantum leap in X-ray spectroscopy has been provided by combining the energy resolution of wavelength-dispersive spectroscopy (WDS) with the ease of use of the energy-dispersive X-ray spectroscopy (EDS) technique. The POLARIS microcalorimeter X-ray detection system from VeriCold Technologies GmbH (Ismaning, Germany) is a novel EDS-type detector for use with SEM, FESEM, FIB and defect review tools. Fields of application include the characterization of submicron particles and feature sizes, and elemental surface analysis for development, production, quality control and failure analysis. It is suited for use with in-line/at-line analytic tools and in the failure analysis lab.
| 4. X-ray spectrum of TiN recorded with
a POLARIS microcalorimeter (blue) and Si(Li) detector
(red). |
The tool has been evaluated at Infineon (Munich) with AMD on a SEM with a Schottky field emission electron. It has been optimized for SEM imaging up to 200,000×, where the electron column can be very sensitive to external disturbance caused by vibrations, etc. The exceptional energy resolution of the POLARIS microcalorimeter system is well within the expected and targeted 15 eV (FWHM) at 1.5 keV aluminum Ka radiation.
| 5. X-ray spectrum of WSi2 recorded
with POLARIS microcalorimeter (blue) and Si(Li) detector
(red). |
The fine energy resolution of the system enables the user to carry out low-keV X-ray microanalysis taking advantage not only of the improved lateral and vertical resolution but also reduced charging and damaging effects. Furthermore, high-collection-efficiency X-ray optics have now been added that reduce the analysis time by almost a factor of 100, bringing analysis times down to several minutes.
In-line SIMS (ILSIMS)A new generation of high-performance secondary ion mass spectroscopy (SIMS) from Cameca (Courbevoie, France), initially targeted at in-line operation, has so far demonstrated suitable at-line performance. It has been evaluated by STMicroelectronics (Crolles, France) with partners Agere Systems, Siemens, Samsung Electronics and ITC-irst. The IMS Wf is a SIMS instrument based on a double-focusing magnetic sector analyzer. It provides the high mass resolution necessary to resolve mass interferences such as 30 SiH/31 P, and the high sensitivity to achieve sub-ppm elemental detection limits for dopants and impurities present in advanced technologies. Major new features incorporated in the tool include a FOUP-compatible 300 mm full-wafer analysis capability, high automation level and ultralow-energy primary ions capability.
The continuous growth in use of SIMS analysis necessitates that SIMS move from outsourcing to in-line or at-line to improve cycle times and reduce cost. To meet this new demand, SIMS instrumentation must offer new features, including the capability to measure patterned product wafers, with accurate and repeatable monitoring of the detection of species at high sensitivity and with good depth resolution, flexibility to handle the evaluation of small samples, 200 and 300 mm wafers, high degrees of automation, and high uptime and MTBF.
Figure 6 shows results measured within a crater sputtered in a 300 × 300 µm region. The overlay of the three profiles demonstrates that, for arsenic analyzed at high impact energy with Cs+, the raster size can be reduced down to 50 µm with no loss in terms of depth resolution and detection limit. Patterns as small as 100 × 100 µm can therefore be considered for analysis.
The ability of SIMS has been demonstrated, being the only technique to measure carbon doping level within a SiGe compound. The vacuum system of the IMS Wf enabled measurement with a detection limit in the high 1017 at/cm3. Excellent performance on shallow implants monitoring was also demonstrated.
The evaluation has demonstrated that the targeted analytical performance has been reached (and even surpassed on many points). The work has led progressively to the definition of improvements that impact the level of automation and equipment reliability. These improvements have been implemented in second-generation equipment that is installed for at-line use at the STMicroelectronics 300 mm facility in Crolles.
New SEA projectsSEA has started several new projects since we last reported our activities in this publication (see "More Results Flow from the SEA," Semiconductor International, October 2001).
Combined X-ray fluorescence and X-ray reflectivity (COMBIMEXX): Evaluation of a new in-line combined XRF and XRR tool from Jordan Valley (Migdal Ha'Emek, Israel) is being done by LETI/ STMicroelectronics, Philips, Infineon and International SEMATECH. The equipment can determine the thickness, roughness and density of very thin layers using XRR, and the thickness and elemental composition of thicker layers using µ-spot XRF. In tandem, these two techniques enable angstrom-level accuracy and resolution for a large thickness range of 1 nm to 5 µm. Use of XRR for in-line control in the fab and reduction of the incident spot size for XRF from 1-4 mm to 30-40 µm should enable X-ray control directly on patterned product wafers for the first time.
Low-energy PECVD hetero-epitaxy (LEPCVD300): Advances in silicon semiconductor technology are restricted by the approach of the material's physical limits. Alternate wafer materials (e.g. GaAs and InP), while having enhanced properties, entail considerably greater manufacturing complexity and cost. Fortunately, germanium and silicon exhibit very similar chemical behavior, so adding germanium to the mix hardly impacts most current CMOS processes. Unaxis Semiconductors (Liechtenstein) has developed a process to accurately deposit SiGe on 300 mm wafers much faster than existing techniques. There is consensus from most advanced IC makers that strained silicon will be implemented for real devices from 2003. LETI/STMicroelectronics, Motorola, Picogiga and Wacker Siltronic are assessing this 200/300 mm industrial tool for economic and reliable fabrication of SiGe pseudo substrates by growing relaxed layers of SiGe on silicon wafers.
Advanced CVD tool for integration of low-k films (ACTION): The move to multilevel copper interconnect dual-damascene schemes for the 90 nm node and below requires new materials with a dielectric constant of 2.2 to reduce RC delays, power consumption and cross-talk between metal lines and levels. Good physio-chemical stability and mechanical integrity through repeated and aggressive thermo and mechanical process cycles while being integrated into multilevel stacks is essential. The equipment being assessed from Trikon Technologies (Newport, UK) by STMicroelectronics, AMD, Philips, International SEMATECH and the Technical University Chemnitz is capable of chemical vapor deposition (CVD) of a dielectric material on 300 mm wafers with a k=2.2 by forming small controlled voids in the organosilicate glass matrix, thus significantly decreasing the dielectric constant of the porous material. The equipment is being evaluated for integration of such low-k materials together with dielectric barriers and liners to sequentially deposit the complete stack for an ultralow-k material capped with a hard mask.
High-throughput ALCVD batch equipment (HALE&RAPID): Generally, ALCVD processing is aimed at all applications 90 nm and beyond where high-k dielectric materials are required for deep trench technologies with aspect ratios of ~60:1, high-k gate oxides, copper-based metalization schemes and for hydrogen barriers in FeRAM technologies. The anticipated processes targeted in this project — by Infineon, Austria Microsystems, Motorola and the FHG IIS-B — are Al2 O3 layers for memory trench capacitors and Ta2O5 layers for metal insulator metal (MIM) capacitors within the metalization schemes for wireless products. The consortium will assess ALCVD with ASM International's (Bilthoven, Netherlands) modular A400 AL, a new member of the high-throughput A400 vertical furnace family. The system provides ALCVD batch processing, remote plasma treatment and in situ chamber cleaning. Key aspects of the evaluation will be optimization of equipment reliability, including cleaning issues, overall uptime, throughput and CoO.
| Author Information |
| Jeff Bruchez has worked in the microelectronics industry for 36 years, holding senior positions in R&D, IC manufacturing and with semiconductor equipment companies. He is now an independent technology consultant, director of SEA dissemination, and responsible for the independent experts who monitor, evaluate and disseminate SEA projects. |
| E-mail: sea@rl.ac.uk. |
| Acknowledgements | ||
| SEA projects are partially funded by the European Community's IST program. The author acknowledges the contributions of all participants in these projects. | ||
|