Measuring the Thickness of Micromachined Silicon Wafers Using Machine Vision
Author(s):
Thomas Digges Jr. - Virginia Semiconductor
Robert A. Ross - Virginia Semiconductor
Industry:
Semiconductor
Products:
Vision, Motion Control, LabVIEW
The Challenge:
Using nondestructive techniques for automated measurement of the thickness of silicon wafers containing micromachined structures.
The Solution:
Developing a PC-based optical micrometer using off-the-shelf tools for data acquisition, image processing, and motion control in order to reduce development time and improve marketability.
"We achieved our goal using LabVIEW, DAQ, IMAQ, and ValueMotion products from National Instruments."
Introduction
Silicon wafers are traditionally thought of as the base material from which electronic integrated circuits are produced. More recently, designers have created a miniature electromechanical structure (MEMS) from this same base material. These miniature structures are etched from the base silicon material, with particular regions mechanically relieved and other regions firmly anchored. As a result of this process, the MEMS device exhibits a particular mechanical action, such as torsion or linear motion. MEMS devices are commonly used as pressure transducers and accelerometers.
For example, a MEMS pressure transducer might consist of one or more micromachined diaphragms measuring 25 µm to 5 mm in diameter, and as thin as 10 µm. Applied pressure produces a deflection in the silicon diaphragm, which produces an electrical signal that can be amplified and calibrated to pounds per square inch (psi).
Virginia Technologies, Inc. was retained by Virginia Semiconductor, Inc. to develop an optical micrometer capable of measuring MEMS device thickness with resolution in the µm range. Our goal was to use off-the-shelf tools for data acquisition, image processing, and motion control in order to reduce development time and improve marketability. We achieved this goal using LabVIEW, DAQ, IMAQ, and ValueMotion products from National Instruments.
System Design
The host computer for the OMMS-1 is a 200 MHz Pentium Pro, running Windows NT, with a total of 160 MB RAM for the extensive image processing required. We use the analog and digital outputs of an AT-MIO-16XE-50 plug-in DAQ board for laser control. Digital inputs on the board are used to monitor the status of safety interlock switches on the lasers and moving components. A PCI-1408 plug-in image acquisition board acquires the wafer images from a charge-coupled device (CCD) camera. We also use a ValueMotion PC-Step-4CX plug-in board to control the XY-positioning stage.
We chose LabVIEW for our application software because with it we could rapidly generate a graphical user interface for display, analysis, and control. The extensive VI libraries of NI-DAQ for data acquisition, NI-IMAQ and IMAQ Vision for image acquisition and processing, and the ValueMotion VI library for motion control provided all the functions required for a complete system.
System Performance
We developed the OMMS-1 (optical micrometer for micromachined substrates) for inspecting double-sided polished (DSP) micromachined substrates. This instrument calculates and displays an absolute thickness map for any region of the wafer by using a broad optical beam as a probe. Direct absolute thickness measurement is accomplished across the area of the beam. Because measurements are continuous over the region, thickness and total thickness variation for individual membranes and etched features can be evaluated accurately, rapidly, and nondestructively.
The OMMS-1 can accurately map the thickness of micromachined features as thin as 10 µm and as small as 100 µm in width. The instrument can measure feature step heights with a resolution of approximately 0.5 µm. The operator positions a wafer riding on an XY-positioning stage to various locations for evaluation using a mouse-driven graphical user interface. Absolute thickness maps are stored in a database for easy process monitoring and tracking.
The OMMS-1 instrument uses lasers whose energies are focused through the silicon wafer under test. The instrument then maps the energy attenuated by the wafer to the wafer thickness. The user defines any number of 0.25 by 0.25 in. regions on the wafer for evaluation. Once a particular region is selected, the XY-positioning stage centers that region under a CCD array. The user then begins the evaluation process by simply pushing a button on the LabVIEW panel.
After the image is acquired and processed, we use the IMAQ Vision VI libraries extensively to process the acquired image and calculate wafer thickness information. A LabVIEW intensity graph displays the thickness of each pixel in the evaluation region as an 8-bit grayscale value. The user can then select vertical, horizontal, or diagonal line profiles showing the thickness value across the evaluation area.
Results
The OMMS-1 is currently being beta-site tested by a manufacturer of MEMS pressure transducers. Testing done to date has shown excellent results in thickness measurement resolution and dynamic range. An estimated 50% reduction in development time and cost was achieved by using a LabVIEW-based solution as compared to a solution based on the development of custom electronic hardware and Visual C++ programming. Availability of the production unit is anticipated in the third quarter of 1998.
Future capabilities we plan for the instrument include options to store data from defined evaluation regions for production test. We also plan to enhance the database capabilities of the instrument to include storage of images, line profiles, and user notes indexed by wafer serial number.
For more information, contact:
Dr. Thomas Digges
Virginia Semiconductor Corporation
1501 Powhatan Street
Fredricksburg, VA 22401
Tel: (540) 373-2900; or
Robert A. Ross
Virginia Technologies, Inc.
2015 Ivy Road, Suite 423
Charlottesville, VA 22903
Tel: (804) 970-2200
Related Case Studies
Automated Silicon Wafer Measurement Using NI-IMAQ and LabVIEWQuantitative analysis of cytoskeletal fiber orientation using IMAQ Vision library
Quantitative Analysis of Cytoskeletal Fiber Orientation
LabVIEW-Controlled Robot Climbs and Inspects Highway Lighting Towers
Automated Automotive Body Inspection System for Poka-Yoke Production Lines Using NI IMAQ Vision
|
|
