Upload
serrot-onaivlis
View
232
Download
1
Embed Size (px)
DESCRIPTION
research on x-ray crystallography
Citation preview
A Study on Parasitic Slit-Scatteringsin SAXS Equipment
Tuo HuangGrinnell College, IA, USA
PI/Mentor: Dr. Youli LiMaterials Research Laboratory, University of California, Santa Barbara, CA, USA
August 2007, University of California, Santa Barbara
I. Introduction
The X-Ray Diffraction (XRD) Facility is one of the most frequently used user-
facilities lab at MRL, UCSB. By using X-ray diffractometers, researchers who work
with nanosystems, such as organic/inorganic/hybrid crystals, biomacromolecules,
semiconductor devices, polymers, microfibers, etc, are able to obtain information
about the nanoscale structures of their materials. Among the many XRD techniques,
Small Angle X-ray Scattering (SAXS) is best suited for examining nanostructures on
the scale of 101 to 102 Å. In SAXS instruments, incident X-ray beam must be well
collimated and have a clear edge, so that user can clearly distinguish the scattering
signals from the transmitted direct beam. To obtain such a beam profile, multiple
apertures formed by pairs of horizontal and vertical slits made of heavy metals such as
tantalum or tungsten, which are strong x-ray attenuators, are used to define the
geometrical shape and size of the beam.1 However, slits made from such
polycrystalline metals, no matter how finely machined and polished, will generate
undesirable "parasitic scattering", where slits themselves scatter X-ray photons in a
wide angle and produce strong background noise in the beam profile.
In this research, we conducted a series of experiments to characterize the slit
scattering from different materials under various conditions in order to find possible
ways to enhance the performance of SAXS instruments.
Page 2 / 11
II. Experiment Approaches
We first studied the scattering profile of traditional tungsten blades from actual
x-ray slits widely used in x-ray diffractometers (Figure 1). One of such slits was
placed at the sample-holder of the SAXS instrument and moved across the x-ray beam
to produce slit-scattering images (the incident beam is normal to the edge of the slit).
Before the image was taken, an X-ray pin-diode was placed right behind the slit to
measure the X-ray beam intensity when the slit was moved to a certain position.
Page 3 / 11
Scattered Beam(To detector)
Incident Beam
Slit moves into the beam
Figure 2 shows that as the slit cuts across the beam (the slit moves in negative x
direction), the X-ray intensity drops because a larger part of the beam is blocked. The
plot above also shows that, on the interval where the intensity is between 5000 and
20000 (counts per second), intensity drops linearly, which means that the slit is
cutting into the central part of the beam. This interval is chosen for data-taking so that
scattering at different cutting-depth happen on the same length-interval on the edge of
the slit.
After taking the intensity-depth plot, the pin-diode is removed, and the slit is
driven to selected points of cutting-depths by high-precision stepping-motors
(accuracy within 10 microns) and a two-dimensional scattering image was taken at
each point with a multiwire x-ray area detector (Bruker HISTAR).
The aforementioned steps were then repeated, only replacing the tungsten slit
with a piece of silicon wafer, with its 1 0 0 crystal plane facing the beam and 0 0 1
plane on the edge (so slit scattering, if any, should occur on the 0 0 1 plane).
Fig. 3: Experiment with Silicon wafer
The scattering profile of silicon wafer shows reduced slit-scattering, but the
scattering from the wafer’s edge still fits into a Lorentzian function. Attempts were
made to polish the (0 0 1) plane for greater flatness but that did not yield much
improvement. However, we noticed that the (1 0 0) plane was very well polished
Page 4 / 11
(1 0 0) Plane(0 0 1) Plane
Incident X-ray Beam
when the wafer was produced; therefore, if we could make (1 0 0) plane instead of (0
0 1) plane on the edge, scattering may be significantly reduced.
This resulted in making a tungsten slit with a tiny strip of silicon wafer mounted
on the edge (Figure 4). This hybrid slit was again tested using the same method as
above.
III. Results and Analysis
1. Tungsten Slit & Single-Crystal Silicon Slit
To explain the analysis of scattering profile, let us take the image of tungsten slit
at position x=-6.17 and intensity I=12827 Counts/sec as an example. All the other
images taken are processed in the same manner.
The image is first displayed from the raw data file (Figure 5). After finding the
pixel position of the beam center (597.2536, 538.6029), a rectangular area is defined
along the X direction i.e. the direction of the scattering “tail.”
Fig. 5: Analyzing the Scattering Tail
Page 5 / 11
Tungsten Slit
Silicon Wafer
1-0-0 Plane
Incident X-ray Beam
Fig. 4: Hybrid
Slit
Then numerical integral is computed along the X axis of the image to reflect the
average intensity of the points in the rectangular at each x position. The resulting plot
is shown in Figure 6. In the plot, x axis is the 2-theta scattering angle that each point
corresponds to (label on x axis is in reverse order, and the peak corresponds to the
scattering “tail” in the image):
1 J. S. Pedersen in Modern Aspects of Small-Angle Scattering edited by H. Brumberger (Kluwer Academic Publishers, Dordrecht/Boston/London, 1995), 57-91
Page 6 / 11
Figure 6: Result of Numerical Integration along the x-axis
The result is then exported to a two-column chiplot file that can be read by
analysis software such as MS Excel™ or Origin™. The data was further assorted and
the data points that are on the scattering “tail” were extracted to be studied alone for
curve fitting.
The scattering profiles of both tungsten and Silicon Wafer slits fit into a
Lorentzian function in the form of where A and Γ are
variables positively correlated to the height and width of the peak, respectively.
Fitting results are shown in Table 1 & 2.2
Page 7 / 11
Table 1: Curve-fitting Results for the Scattering Profile of Tungsten Slits
Relative Slit
Position (mm)
y0 x0 ΓError for
Γ(±)A
Error for A (±)
Goodness of Fit
Reduced χ-square
R-square
0.1 0.37544 -0.10555 0.08392 0.00062 2.55588 0.02912 0.0212 0.997610.3 0.14054 -0.10711 0.08135 0.00051 6.57875 0.06011 0.1095 0.998280.5 0 -0.10576 0.07709 0.00046 14.48919 0.13391 0.60831 0.998080.7 0 -0.1055 0.07277 0.00046 19.62139 0.19278 1.28709 0.99790.9 0 -0.1057 0.07012 0.00047 18.51033 0.19293 1.36157 0.997621.1 0 -0.10615 0.06815 0.00051 11.93119 0.13408 0.71214 0.99715
Table 2: Curve-fitting Results for the Scattering Profile of Silicon-
Wafer Slits
Relative Slit
Position (mm)
y0 x0 ΓError for
Γ(±)A
Error for A (±)
Goodness of Fit
Reduced χ-square
R-square
0.1 0.44246 -0.07578 0.06998 0.00239 1.59566 0.11902 0.0071 0.992590.3 0.3661 -0.06647 0.06178 0.00413 3.28324 0.33896 0.01076 0.995170.5 0.24839 -0.07342 0.06174 0.00171 4.58296 0.2264 0.01238 0.998050.7 0.11366 -0.07905 0.05567 0.00113 5.71605 0.22145 0.0195 0.998490.9 0.01246 -0.08087 0.04189 0.00171 6.36557 0.4097 0.03271 0.997721.1 0 -0.08774 0.03415 0.00081 4.61314 0.17914 0.01491 0.99871
2. Hybrid Slit
Numerical integration was also done for the images from the hybrid slit and
compared with the background (Figure 7). The negative half on the horizontal axis
corresponds to the scattering tail on the image. The profile of the hybrid slits (blue
curves) are basically superposed on the background (dark-red curve) i.e. having
almost the same intensity with the background at each point, but only lower because
2 In curve-fitting, y0 is forced to be zero if the automated fit returns a negative y0 value, because the background offset cannot possibly be negative in this case.
Page 8 / 11
the overall intensity decreases as the slit moves from x= -6.2 to x= -6.6, blocking a
greater part of the beam. The profile from the hybrid slit is insignificant compared to
scattering from tungsten slits (gold curve).3 Hence, we concluded that the hybrid slit
has no obvious slit-scattering.
IV. Discussion
We are not sure about what has caused the Lorentzian pattern in the scattering
tail, which can also be modeled by a -2 power function. S.K. Sinha et al.4 suggest that
it may be due to the exponentially distributed roughness of the surface and, in the case
of tungsten, also the randomly oriented polycrystalline micro-domains.
3 Data from tungsten slit is not extracted from the ones analyzed above, but was retaken in the same set-up as the hybrid slit right after the hybrid slit’s data was taken in order to show more valid comparison.4 S. K. Sinha, E. B. Sirota, and S. Garoff. Phys. Rev. B 38, 2297 - 2311 (1988)
Page 9 / 11
The absence of slit-scattering for hybrid slit is both exciting and well within
expectation. Because the (1 0 0) plane is well polished and has no roughness, when
this plane is on the edge of the slit, the lowest detectable x-ray diffraction angle is the
one at 69.13 degrees for the (4 0 0) plane (which is parallel to the (1 0 0) plane).5 Yet
in SAXS, the X-ray beam is collimated to a divergence angle of less than 0.05 degrees
and the detection range of scattered angle is less than 2 degrees, both of which
characteristics make it impossible for any X-ray photons scattered at a 69.13-degree
angle to be picked up by the detector.
V. Conclusion
After studies on tungsten and silicon wafer slits, we can make a tentative
conclusion that roughness of the surface on the edge of the slit seems to be the
primary cause of slit-scattering. Meanwhile, scatterings from polycrystalline micro-
domains also contributed to the overall slit-scattering and can be reduced by using
single-crystal material for slits. When both factors are taken care of, as in the case of a
hybrid slit, slit-scattering is reduced to undetectable level.
If further experiments continue to show satisfactory results, these newly
designed hybrid slits can lead to dramatic improvement to optical configuration of the
SAXS instrumentation. In the traditional three-slit setup, besides the two set of slits
that collimate the X-ray beam, additional slits are needed to cut off slit-scattering
from those previous slits. In contrast, hybrid slits do not generate slit-scattering and
thus can significantly simplify the design of SAXS equipment and boost the
5 Natl. Bur. Stand. (U.S.) Monogr. 25, 13, 35, (1976)
Page 10 / 11
efficiency and performance of the equipment.
VI. Acknowledgements
It is a pleasure to acknowledge the superb technical assistance from Joanna
Deek, Nicholas Judy and Eric Welsh, and the help and guidance from the program’s
coordinator, Dr. Patricia Halpin and supermentor Brett Brotherton.
This work was supported by the MRSEC Program of the National Science
Foundation under Award No.DMR05-20415.
Notes and Citations
Page 11 / 11