Electron microscopes are limited to biological specimens no thicker
than a bacterium, and the technological challenges involved with
making high-efficiency diffractive x-ray optics limits the capabilities
of current lens-based x-ray microscopes. Coherent x-ray diffraction
imaging, also known as x-ray diffraction microscopy (XDM), uses
only the scattered intensities in the far field to recover the structure
of the scattering object. If the object is nonperiodic, then the
far-field intensity pattern is continuous and can be sampled finely
enough that the phase, which is lost when intensity measurements
are made, can be recovered by iterative computational methods (see
previous highlight, "Demonstration
of Coherent X-Ray Diffraction Imaging").
Experimental diffraction data used as input to
the difference map algorithm. The continuous diffraction pattern,
obtained after approximately 60 s of exposure to the x-ray beam,
extends to 10-nm resolution at the edges and spans over six decades
of intensity. Insets show magnified views of boxed regions.
In XDM, the action of a lens is replaced by a computational process
that recovers the far-field phase information through the iterative
application of constraints in both image and diffraction space.
These constraints describe a priori knowledge that we have about
the object and typically take the form of a support constraint (e.g.,
a requirement that the sample be finite and isolated) and a Fourier
modulus constraint (a requirement that the calculated diffraction
intensities match the measurements). This method of imaging a non-periodic
object by phasing its continuous diffraction pattern was first suggested
by David Sayre (then at IBM T.J. Watson Research Center) and first
demonstrated with x rays by John Miao (then at Stony Brook University)
and others in experiments at the National Synchrotron Light Source.
The current work represents the first application of XDM to image
an object as complex as a eukaryotic cell. The diffraction microscope,
developed by researchers at Stony Brook and now stationed at ALS
Beamline
9.0.1, is capable of collecting three-dimensional diffraction
data sets from dry or frozen hydrated specimens to a scattering
angle that corresponds to a reconstructed half-period pixel size
of 6 nm for 750-eV x rays. The diffraction pattern that the researchers
used for their image reconstruction is a subset (1200 x 1200 pixels)
of the full CCD recording and extends to 10-nm resolution. Approximately
400 intensities at low spatial frequencies are missing because of
the need to block the brightest part of the pattern to prevent damaging
the detector.
(A) Reconstruction of the x-ray wavefield after
passage through the cell. The phase information is represented
as hue (see color map at left) and the magnitude as brightness.
The labels indicate the nucleus (N), vacuole (V), and the cell
membrane (M). (B) Scanning transmission x-ray microscope image
of the same cell (obtained at the National Synchrotron Light Source
with 540-eV x rays and an optic with a Raleigh resolution of 42
nm). This image is for comparison only and was not used in the
reconstruction process. (C–D) Two independent reconstructions
after the cell was rotated by 3 and 4 degrees with respect to
that in (A). The insets show that similar structures on the scale
of 30 nm can be clearly seen in each.
The phasing algorithm, developed at Cornell and known as the difference
map, controls these intensities and calculates the missing phases.
The reconstructed image contains both magnitude and phase information
(not the same as the diffraction phases calculated by the algorithm)
about the scattered wavefield. An averaging technique was developed
to minimize the effects of noise on the reconstruction. Similar
reconstructions were obtained from eight angular orientations of
the cell at 1° rotation intervals. The good agreement between
the independently recovered structures provides confidence in the
fidelity of the reconstructed images, and a comparison of adjacent
reconstructions indicates a spatial resolution of better than 30
nm. This estimate is supported by an analysis of an effective modulation
transfer function for the microscope.
Experiments aimed at imaging a frozen hydrated cell in three dimensions
are ongoing and will allow us to test our calculations that show
the radiation-damage-limited resolution of XDM to be 10 nm for biological
specimens.
Research conducted by D. Shapiro, E. Lima, H. Miao, A.M. Neiman,
and D. Sayre (Stony Brook University); P. Thibault and V. Elser
(Cornell University); T. Beetz and C. Jacobsen (Stony Brook University
and Brookhaven National Laboratory); M. Howells (ALS); and J. Kirz
(Stony Brook University and ALS).
Research funding: National Institutes of Health, National Science
Foundation, and U.S. Department of Energy, Office of Basic Energy
Sciences (BES). Operation of the ALS is supported by BES.
Publication about this research: D. Shapiro, P. Thibault, T. Beetz,
V. Elser, M. Howells, C. Jacobsen, J. Kirz, E. Lima, H. Miao, A.M.
Neiman, and D. Sayre, "Biological imaging by soft x-ray diffraction
microscopy," Proc. Nat. Acad. Sci. USA 102,
15343 (2005). |