wafer thickness ttv bow and warp for thin wafer ... · pdf file1 by thomas bristow wafer...
TRANSCRIPT
1
by
Thomas Bristow
Wafer Thickness TTV Bow and Warp for Thin Wafer Application
For SEMATECH Workshop on 3D Interconnect Metrology
July 11, 2012
Proprietary and Confidential Information 2
MPT1000 Wafer Measurement System
Comprehensive Non-contact
Measurement System• Thickness
• Total Thickness Variation
• Bow
• Warp
• Surface Roughness
• Edge Chip and Crack
• Tape Thickness
Proprietary and Confidential Information 3
MPT1000 Specifications•Thickness Resolution < 0.1µm
•Repeatability 0.1 µm
•Accuracy 0.15 µm
•Range 10 µm-12 mm
•Wafer Size Up to 300 mm
•Safety Semi S2/S8
Abstract
Several metrology avenues are required for 3D Interconnects. One of the features of these products is the drive for thinner wafers. The
metrology requirements for thinner wafers include wafer thickness, TTV, surface roughness and bow and warp. Bow and warp metrology
require additional definition for the support of the wafer during the measurement. Wafer metrology for bow and warp of un-patterned wafers follow the SEMI definition of supporting the wafer on three
small balls near the wafer edge and measuring both the front and back side (flipping the wafer). This method has the advantage of
minimizing gravity effects. For thin wafer metrology this approach has limitations in the wafer handling and support during the measurement. This paper will present several methods for wafer support for bow and
warp measurements.
How to best measure bow and Warp?
It’s not as straightforward as TTV to measure bow and warp.
TTV gives the maximum thickness variation independent of how the wafer is held.
Bow and warp are influenced not only by the wafer stress, but how the wafer is held and by gravity.
SEMI standards are available for background on bow and warp measurements.
Thin bowed wafers represent some special challenges for bow and warp measurements.
There are three generally accepted methods for measuring bow and warp, each with an advantage and disadvantage.
Method 1 (the most common) has the wafer positioned flat on the chuck (sometimes called Sori)
Method 2 (one of the SEMI standard methods) has the wafer positioned on three small balls. The advantage to this method is that no part of the wafer is touching the chuck and thus the chuck does
not influence the measurement. The disadvantage is that gravity strongly influences the result.
Method 3 (one of the SEMI standard methods) has the wafer positioned on the chuck and measures both top and bottom of the
wafer. This method has the advantage of minimizing gravity effects.
We want to measure warp because the wafer stress is directly proportional to the warp.
Warp is the difference between the maximum and minimum distances of the median
surface of a free, unclamped wafer from a reference plane.
A reference plane is derived from a least square calculation of values of the warp measurement.
The figure below shows an example of a warp wafer with maximum distances above and below the
reference plane.
Warp calculation
Warp = MaxD – MinD
Where MaxD = maximum distance above the reference plane and
MinD = maximum distance below the reference plane.
The figure below shows a wafer supported by three small balls near the wafer edge
Warp results effected by:
Method of holding the wafer
The following examples show warp on two types of wafers, a low warp (thick 725µm) and a high warp (thickness =273µm) wafer.
Each wafer measured by three methods:
Measure both sides of the wafer and subtract the two results.
Measure just the front side of the wafer
Measure the warp on just the front with the wafer supported on three small balls near the wafer edge.
Theoretical Considerations.
Stress goes as 1/R where R is the radius of curvature of the wafer
R=D2/8W
Where D= wafer diameter and W= warp
Gravity can be a significant influence on the warp especially for thin wafers
The sag at the wafer edge S is give by
S=KD4/t2
where K= 7.83x10-3 um3/mm
D=wafer diameter and t= wafer thickness
Examples show good correlation using gravity correction and difference from front only vs. front-back measurements
S=23.8µm for D=200mm and t=725µm
S=168µm for D=200 and t=273µm
Method Summary for high warp wafers.
Use method of:
Wafer flat on the chuck (sometimes called Sori)
Measure the warp from one side and compare with other wafers of the same type
For an accurate stress measurement it’s necessary to subtract out gravity effects, either by using a reference wafer and
measuring both sides, or a theoretical model to subtract out the gravity effects.
Other Considerations
Warp results effected by:
Diameter and thickness of the wafer
Gravity compensation
Method of holding the wafer
Number of Data Points
Measuring instrument resolution and range (range should be greater than several mm for high warp wafers)
Repeatability and Reproducibility should be measured
Following is an example of reproducibility by comparing the wafer warp contour plot after rotating the wafer
Proprietary and Confidential Information 21
Conclusions• Warp Measurements for thin wafers influenced by several
factors– Wafer Diameter and thickness– Gravity considerations– Method of holding the wafer– Number of Measurement Points
• Best method for production applications is to hold the wafer flat on a chuck and compare results with similar wafers
• Comparison with other tools and/or companies best done with the same methods.
• Accurate Stress can be calculated after removing gravity effects.
• References– SEMI MF1390-0707– SEMI MF657-0707E
THANK YOU!Chapman Instruments
3 Townline Circle Rochester, NY 14623
Phone: (585) 272-9994 Fax: (585) 272-9996
Email: [email protected]: www.chapinst.com