Browse Prior Art Database

Phase-Sensitive Overlay Analysis Spectrometry

IP.com Disclosure Number: IPCOM000099986D
Original Publication Date: 1990-Mar-01
Included in the Prior Art Database: 2005-Mar-15
Document File: 5 page(s) / 154K

Publishing Venue

IBM

Related People

Korth, HE: AUTHOR

Abstract

Accurate measurements of the overlay between structures on, say, semiconductor wafers may be made by spectrometrically analyzing sets of superimposed gratings. An algorithm is described which evaluates the symmetric properties of the gratings. This allows offset measurements that are largely independent of grating profile distortions.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 47% of the total text.

Phase-Sensitive Overlay Analysis Spectrometry

       Accurate measurements of the overlay between structures
on, say, semiconductor wafers may be made by spectrometrically
analyzing sets of superimposed gratings.  An algorithm is described
which evaluates the symmetric properties of the gratings.  This
allows offset measurements that are largely independent of grating
profile distortions.

      Computer simulation indicates that an accuracy of some 3% of
the overlay value + 1 nm can be maintained.

      Overlay Spectrometry For the analysis of grating pattern depth,
computer-based white- light spectrometry may be used. Well-known film
thickness algorithms are available for evaluating the interference
pattern of the zeroth order of diffraction of the grating.  Such an
interference fringe pattern is produced if two grating structures are
not perfectly overlapping.  If one of the gratings is shifted by half
a grating period, another fringe pattern results.  The transition
between these patterns is a function of the offset between the
gratings.

      An example of an overlay measurement set-up is shown in Fig. 1.
 A group of four grating patterns with a diameter of some 10 mm for
each direction is printed with the features whose overlay is to be
tested.  The spectral reflectance of the composite pattern can then
be sequentially sensed. Parallel sensing may be done with a multiple
spectrophotometer generating four spectra along a linear photodiode
array.  This allows exploiting the larger number of photodiodes (say,
256) of current diode array products, with the moderate spectral
resolution requiring less than 50 signal channels per spectrum.

      Model Calculation The difference between the signal from a
sine-shaped pattern and the signal Fkt(x) from a trapezoid was
calculated for various values (Fig. 2).  This difference defines the
maximum overlay error that may arise if only limited information is
available on the real signal shape.  The grating period should be
selected trading the resolution of the photolithographic process
against signal noise.  A 2 mm grating period may be a good value for
a process with a resolution better than 1 mm.

      Three scenarios are discussed below based on the data of Fig.
2.
      1.   There is no information on the shape of the superimposed
gratings, i.e., the linewidth may have a value of between 600 and
1400 nm for one grating and 800 and 1200 nm for the other. The
deviation may reach a maximum of 0.08 corresponding to an offset
error of about 20 nm. Below 10 nm offset, the maximum deviation may
reach some 20% of the offset value.  The mean value between
x/(0.25-x) and tan(2fx) may be used to decode the offset.  This
reduces the maximum error to 10%.
      2.   One grating is processed to appear with u = 1/f to the
meas  uring system.  u may vary within a range of 5%.  This means a
linewidth of one grating of 600 to 660 nm.  The linewidth of the...