Hawkes P.W., Spence J.C.H. (Eds.) Science of Microscopy. V.1 and 2

Подождите немного. Документ загружается.

888 M. Howells et al.

the depth of focus. The associated questions of what will be the cost

in resolution, contrast and efﬁ ciency are also now beginning to be

answered favorably. This approach has been used to obtain sub-100 nm

resolution tomographic reconstructions of metallic layers within

thinned integrated circuits using a laboratory X-ray microscope operat-

ing at 5.4 keV (Wang, 2002) as will be described in Section 4.3. For lower

density specimens, the use of hard X-rays naturally leads to the use of

phase contrast which is far more dose-efﬁ cient than absortion contrast

in this energy region. As shown in Figure 13–5, this enables multi-keV

imaging at similar dose levels to the water window. As noted in Section

3.3.7 there have already been quite successful demonstrations of phase

contrast tomography using hard X-rays (Cloetens et al., 1999) in the

past though not (to our knowledge) at the sub-100 nm resolution level

accessible to zone plate microscopes. Now the group at NSRRC, Taiwan

have recently used their 8 KeV phase-contrast TXM with a 50 nm zone

plate to produce 3D images of a microcircuit with defects at 60 nm

resolution (Yin et al., 2006) (see Sections 1.4 and 2.4.4).

3.4.7 Avoiding the Depth-of-Focus Limit by Lens-Free Imaging

The challenges of achieving the highest possible resolution in 3D

imaging have led to the consideration of lens-free imaging. Of course

crystallography is able to obtain exquisite 3D maps of the electron

density of a unit cell in a crystal by interpretation of a tilt series of

Bragg diffraction patterns. In the case of a noncrystalline specimen,

one obtains a continuous rather than Bragg-sampled diffraction pattern

but there has been considerable recent progress in obtaining X-ray

images through the application of iterative phasing algorithms to dif-

fraction data from objects known to be limited in size (Miao et al.,

2002; Marchesini et al., 2003; Williams et al., 2003; Shapiro et al., 005;

Chapman et al., 2006). At the moment the data collection time in 3D

experiments of this type is 10–20 hours and 10 nm resolution images

of materials-science samples in 3D and 30-nm-resolution images of

biological specimens in 2D have been obtained. Moreover, a modern

beam line designed speciﬁ cally for this type of experiment would

easily bring image acquisition times down to a convenient level. It is

noteworthy that the phasing algorithms depend on use of the Born

approximation which sets an upper limit to the sample size which may

become signiﬁ cant at low X-ray energies such as those in the water

window.

3.5 X-Ray Spectromicroscopy

As noted in Section 1, X-ray absorption edges arise when the X-ray

photon reaches the threshold energy needed to completely remove an

electron from an inner-shell orbital. At photon energies within about

10 eV of the edge, electrons can also be promoted to unoccupied or

partially occupied molecular orbitals (see Figure 13–26); photons over

a narrow energy range are sometimes able to excite inner-shell elec-

trons into such orbitals, giving rise to absorption resonances. This so-

called X-ray absorption near-edge structure (XANES) or near-edge

X-ray absorption ﬁ ne structure (NEXAFS) is highly sensitive to the

Chapter 13 Principles and Applications of Zone Plate X-Ray Microscopes 889

local chemical bonding state of the atom in question (Stöhr, 1992) (see

Figure 13–26).

One can exploit these resonances as an additional contrast mecha-

nism in soft X-ray imaging. In electron energy loss spectroscopy (EELS),

the equivalent contrast mechanism is known as ELNES for energy-

loss near-edge structure and its use in energy-loss spectrum imaging

(Jeanguillaume and Colliex, 1989; Hunt and Williams, 1991) is described

elsewhere in this volume. Early efforts in X-ray imaging included the

use of XANES resonances to enhance the sensitivity of differential

absorption measurements of calcium in bone (Kenney et al., 1985),

spectral imaging (King et al., 1989) and microspectroscopy in photo-

electron microscopes (Harp et al., 1990), and photoelectron and trans-

mission imaging at selected photon energies (Ade et al., 1990b, 1992).

It is now common to take image sequences across X-ray absorption

edges (Jacobsen et al., 2000b) yielding data sets with a full near-edge

spectrum per pixel. When comparing spectrum imaging in electron

versus X-ray microscopes, a few comments are in order:

• ELNES is typically done using a ﬁ xed electron energy in the range

80–200 keV. The ideal specimen thickness is under 100 nm in most

cases.

• In ELNES, one gets spectroscopic information over a wide range of

energies, including plasmon energies of ∼10 eV, i n a single measure-

ment. However, plural inelastic scattering dominates the signal at

higher energies (for example, electrons can lose 300 eV once, or 50 eV

six times, etc.) resulting in poorer signal-to-background.

• In X-ray absorption spectroscopy, one must tune the incident

X-ray energy across each absorption edge of interest. The optimum

n=1

n=2

n=3

molecular orbital

Continuum

(fully ionized)

Absorption

Photon energy

Figure 13–26. Schematic of an X-ray absorption edge, which involves the

removal of an inner-shell electron, and a near-edge absorption resonance in

which the electron is promoted to a partially occupied or vacant molecular

orbital. These resonances are referred to as X-ray absorption near-edge struc-

ture (XANES) or near-edge X-ray absorption ﬁ ne structure (NEXAFS).

890 M. Howells et al.

specimen thickness of about 1/µ(E) changes accordingly, so that in

the ideal case one would require samples of several different thick-

nesses to study chemical speciation of several elements. However,

X-rays suffer almost no plural inelastic scattering, which leads to

improved signal-to-background.

• It is common to ﬁ nd scanning X-ray microscopes operating with

monochromators with an energy resolution of 0.1 eV or better. Most

electron microscopes have an energy resolution of 0.5–0.7 eV which

leads to “blurring” of near-edge spectral features, although a limited

number of higher energy resolution systems are starting to become

available.

• Using XANES, one can exploit the favorable characteristics of X-ray

microscopes including the ability to study hydrated specimens and/

or specimens in an ambient atmosphere environment.

In X-ray microscopes, we obtain images (maps of transmitted ﬂ ux I)

according to the Lambert-Beer law for absorption: I = I

exp(−µt) where

is the incident X-ray ﬂ ux, µ is an absorption coefﬁ cient for a speciﬁ c

material, as discussed in Section 1.1, and t is thickness of that material.

The value of µ(E) for near-edge absorption resonances can be calcu-

lated based on the electronic structure of speciﬁ c molecules, and this

has been employed in detailed studies via microscopy of the absorp-

tion spectra of polymers (Urquhart and Ade, 2002; Dhez et al., 2003)

and amino acids (Kaznacheyev et al., 2002) (see Figure 13–27), to name

two recent examples.

For a thickness t of a single material, a measurement of the transmit-

ted ﬂ ux I(E) relative to the incident ﬂ ux I

(E) provides a means to cal-

culate the energy-dependent optical density D(E) = −ln(I(E)/I

(E)) =

µ(E)t. If, however, we measure the optical density not over a continuous

energy range E but at some set of n = 1 . . . N discrete energies E

, we

then measure

= µ

for each of the n = 1 . . . N photon energies. Let us next consider a

mixture of s = 1 . . . S different materials with partial thicknesses t

; our

Arginine:

C=N

Alanine:

aliphatic

Cysteine: side chain -SH

Glutamine: -NH2

Tyrosine: aromatic

292291290289288287

Photon Energy (eV)

292291290289288287 292290288286284

Mass Absorption Coefficient (10

/g)

292291290289288287 292291290289288287

Figure 13–27. Near-carbon-edge absorption spectra of several amino acids, showing the effects of

various molecular bonds in the absorption spectrum. These resonances can be used for chemical

American Chemical Society.) (See color plate.)

Chapter 13 Principles and Applications of Zone Plate X-Ray Microscopes 891

total measurement of optical density D

at one photon energy is given

by the combined absorption of all the materials, or

= µ

+ µ

. . .

+ µ

Finally, if we carry out this measurement not from a single homoge-

neous uniform ﬁ lm, but from heterogeneous pixels p = 1 . . . P indexed

by p = i

column

+ (i

row

− 1) ⋅ (# columns) in an image, the optical density

measured at one pixel p is given by

= µ

+ µ

. . .

+ µ

When all N photon energies are considered, we see that we have a data

matrix D

NNP

NNS

11 1

...

 

























µµ



⋅













SSP

11 1

...



or D

N×P

= µ

N×S

⋅ t

S×P

. In other words the data represent a series of spectral

signatures µ

N×S

and thickness maps t

S×P

When we acquire a series of images at different photon energies N,

we are in fact measuring the data matrix D

N×P

. If we know the exact

absorption spectrum µ

for each of the s = 1 . . . S components in the

sample, then we can ﬁ nd the spatially resolved thicknesses t

S×P

of the

components by matrix inversion:

S×P

= µ

−1

S×N

⋅ D

N×P

The inversion of the matrix of spectra from all known components

N×S

can be accomplished in a robust fashion using singular value

decomposition (Zhang et al., 1996; Koprinarov et al., 2002). This

approach, as well as approaches which involve pixel-by-pixel least

squares ﬁ ts of all reference spectra, work well with specimens that

involve mixtures of components that can all be measured separately.

Examples using this approach are shown in the chemical imaging

section of this chapter.

In many areas of research, such as biology or environmental science,

the complexity of the specimen and the possibility of reactions between

components means that one cannot know in advance the set of all

absorption spectra µ

N×S

present in the specimen. In this case, one

approach that has yielded recent success is to ﬁ rst use principal com-

ponent analysis (King et al., 1989; Osanna and Jacobsen, 2000) to

orthogonalize and noise-reduce the data matrix D

N×P

, and then use

cluster analysis (a method of unsupervised pattern recognition) to

group pixels together based on similarity of spectral signatures (Lerotic

et al., 2004, 2005) (see Figure 13–34 below). This method yields a set of

absorption spectra µ

N×S

where S now indexes the set of characteristic

spectra found from the data. The power of this approach lies in its

ability to improve the signal-to-noise of spectra of heterogeneous speci-

mens by averaging noncontiguous pixels, to ﬁ nd even quite small

regions with distinct spectroscopic signatures, and to deliver continu-

ous “thickness” maps based on the distribution of the discovered sig-

nature spectra.

892 M. Howells et al.

For studies at the carbon edge, one can characterize the observed set

of near-edge resonances in terms of a limited number of functional

group types (see e.g., Scheinost et al., 2001). While there are a number

of open questions regarding this approach (for example, how many

resonances should be used, with what range of allowed center photon

energies, and what range of energy widths?), conﬁ dence in it can be

enhanced by correlation with other spectroscopies such as solid-state

nuclear magnetic resonance (Scheinost et al., 2001; Schumacher et al.,

2005) and Fourier transform infrared (Solomon et al., 2005).

4 Applications

Two decades ago, nearly all research using X-ray microscopes was done

by the groups that had developed the instruments. Today, most X-ray

microscopes are operated as user facilities at synchrotron radiation

research centers, and are used both by their developers but also by a

wider community of scientists. As a result, while it was originally pos-

sible to see the major applications of X-ray microscopes in conference

proceedings (Schmahl and Rudolph, 1984a; Sayre et al., 1988; Michette

et al., 1992), papers in which X-ray microscopes were used to address

the problem of interest now appear across a very wide array of scien-

tiﬁ c journals. In what follows, we do not presume to be exhaustive in

coverage of all research using X-ray microscopes; instead, we will

brieﬂ y highlight a few examples from some of the areas of present

activity.

4.1 Biology

X-ray microscopes using zone plates and synchrotron radiation have

been used for studies of biological specimens from the start (Niemann

et al., 1976; Rarback et al., 1980), and a number of reviews have concen-

trated on biological applications of X-ray microscopes (see for example

Kirz et al., 1995) for background information and older results, or

Abraham-Peskir, (2000). One emphasis has been on high resolution

imaging of whole cells at “water window” wavelengths (see Figure

13–28), including studies of human sperm (Chantler and Abraham-

Peskir, 2004), malaria in red blood cells (Magowan et al., 1997),

Kupffer cells (Scharf and Schneider, 1999) and COS cells

(Yamamoto et al., 1998) from liver, protists (Abraham-Peskir, 1998), and

chromosomes (Guttmann et al., 1992; Williams et al., 1993; Kinjo et al.,

1994) among other examples. As soft X-ray microscopes push to higher

spatial resolution, views through whole cells will involve a great deal

of overlap of structure, but several developments offer information

beyond two-dimensional images with natural contrast. One of these is

to use molecular labeling methods to tag speciﬁ c proteins (such as is

done with great success in visible light microscopy). Several groups

have demonstrated the use of gold labeling in X-ray microscopes,

including detection by dark ﬁ eld (Chapman et al., 1996c) (see Figure

13–29) and bright ﬁ eld (Meyer-Ilse, 2001 et al.,; Vogt et al., 2001b)

approaches. One of the challenges faced thus far is that the label must

Chapter 13 Principles and Applications of Zone Plate X-Ray Microscopes 893

be comparable in size to the resolution of the microscope for efﬁ cient

detection (Chapman et al., 1996c; Vogt et al., 2001a), which means that

in all studies carried out thus far the cell membrane has been permea-

bilized by agents such as methanol to allow relatively large labels to

reach the cell’s interior and this step must be preceded by chemical

ﬁ xation. As a result, future improvements in X-ray microscope reso-

lution will not only lead to improved visualization of unlabeled

7 µm2 µm

Figure 13–28. Whole ﬁ broblast imaged in the frozen-hydrated state. The cell

was cultured on a formvar-coated gold electron microscope grid, and rapidly

frozen by plunging into liquid ethane. It was then imaged using a cryo STXM

operated at 516 eV. In addition to this 2D image, 3D reconstructions were also

obtained using tomography (Wang et al., 2000). (Reprinted from Maser et al.,

Figure 13–29. Human ﬁ broblast with immunogold labeling for tubulin. This

is a composite of two images: a bright ﬁ eld image (gray tones) to image overall

mass, and a dark ﬁ eld image (red tones) to selectively imaging the silver-

enhanced gold labels. This whole-mount cell was ﬁ xed and then permeabo-

lized to allow for introduction of the immunogold labels, after which it was

of America.) (See color plate.)

894 M. Howells et al.

ultrastructure but will also make it possible to use smaller immunola-

bels with more “natural” preparation protocols.

Another approach to exploit the characteristics of X-ray microscopes

is to go beyond two-dimensional imaging. One approach is to use

XANES spectromicroscopy for mapping chemical speciation in bacte-

ria and cells (Ito et al., 1996; Zhang et al., 1996; Lerotic et al., 2005) and

biomaterials (Hitchcock et al., 2002) using the approaches outlined in

Section 3.6 above. Another involves the use of tomography as has been

discussed in Section 3.5. This was ﬁ rst used to study algae in a thin

capillary by Weiss et al. (2000) (Figure 13–30) and to study whole-

mount eukaryotic cells by Wang et al. (2000), followed by studies of

yeast in capillaries (Larabell et al., 2004) (see Figure 13–31). In all of

these cases, cells were studied in the frozen hydrated state for reasons

that will be discussed in the following paragraph. A third approach

beyond two-dimensional imaging is to use X-ray microscopes (Kenney

et al., 1985; De Stasio et al., 1996; Buckley etal., 1997) or X-ray micro-

probes (Kawai et al., 2001; Ortega et al., 2003; Paunesku et al., 2003;

Kemner et al., 2004; Behets et al., 2005; Wagner et al., 2005) to study

elemental content and distribution in bacteria and cells, particularly in

the case of calcium in bone and metals that regulate various biological

activities.

When studying biological specimens, attention must be paid to the

limitations set by radiation damage. Basic considerations of signal-to-

noise and absorption indicate that the radiation dose that is necessarily

imparted for X-ray imaging at 50 nm or better resolution is in excess of

Gray (Sayre et al., 1977a; Schneider, 1998). This is well in excess of

2 µm

Abs. coeff.

µ (µm

-1

)

0.5

Flagella

Flagellar roots

and neuromotor

Nuclear

membrane

Cell wall

Nucleolus

Figure 13–30. 3D rendering (left) and reconstruction slices (right) of the algae Chlamydomonas rein-

hardtii viewed by soft X-ray tomography at the BESSY I synchrotron. This alga was plunge-frozen in

liquid ethane, and imaged over 180º rotation sequence. The reconstruction is given in terms of the

permission from Elsevier.) (See color plate.)

Chapter 13 Principles and Applications of Zone Plate X-Ray Microscopes 895

the <10 Gray (1 Gray = 100 rad) dose that is lethal to humans when

received over a short time interval. Studies of intially living cells have

shown that doses of 10

Gray are at the approximate threshold for

producing immediate changes in bacteria and are well above the dose

needed to affect more complex cells in X-ray microscopy investigations

(Gilbert et al., 1992; Pine and Gilbert, 1992; Bennett et al., 1993; Kirz

et al., 1995). One of the main damage mechanisms is the creation of

radiolytical free radicals in water. Some but not all chemically ﬁ xed,

hydrated biological specimens will show effects such as mass loss,

shrinkage, and the loss of ultrastructural information at these radiation

doses as well (Ford et al., 1992; Williams et al., 1993) (of course, chemi-

cal ﬁ xation produces its own changes on many specimens (Coetzee

and van der Merwe, 1984, 1989; Stead et al., 1992; O’Toole et al., 1993;

Jearanaikoon and Abraham-Peskir, 2005). Fortunately, a ready solution

was developed some years ago by electron microscopists: the use of

rapidly frozen specimens in cryomicroscopy (Taylor and Glaeser, 1976;

Steinbrecht and Zierold, 1987; Echlin, 1992). In X-ray microscopes,

frozen hydrated biological specimens have been shown (Schneider,

1998; Maser et al., 2000) to be well preserved and free of easily visible

structural changes and mass loss at radiation doses up to about 10

Gray thus providing the required conditions for a variety of biological

studies. The situation for spectroscopy is not yet so clear; cryo methods

have been shown to be less effective in preserving XANES resonances

in dry polymers (Beetz and Jacobsen, 2003) but they may be more

advantageous in studies of frozen hydrated organic specimens due

to the inactivation of the diffusion of free radicals (“cage” effect)

(Schneider, 1998).

0.5 µm

Figure 13–31. Single projection image (left) and slice from a tomographic

reconstruction (right) of a frozen hydrated yeast Saccharomyces cerevisiae. A

number of cells were loaded into a thin-walled, 10 µm diameter glass capillary

and rapidly frozen using a jet of helium gas cooled by liquid nitrogen. A series

of 45 images through a 180º tilt range was then acquired using the XM-1 TXM

at the Advanced Light Source. This illustrates the ability of soft X-ray tomog-

raphy to image the interior detail of cells rapidly frozen from a living state.

(From Larabell et al., 2004. Reprinted from Molecular Biology of the Cell, with

permission of the American Society for Cell Biology.)

896 M. Howells et al.

4.2 Environmental Science

Environmental science using synchrotron radiation is a broad topic, as

discussed in a recent review (Brown, 2002); we point out here just a

few examples using X-ray microscopes.

By placing microliter drops between two silicon nitride windows

which are then drawn together by surface tension and some sort of

seal, it is straightforward to make a specimen chamber with microme-

ter-thick water layers and study samples wet and at room temperature

(Neuhäusler et al., 2000) (see Figure 13–32). Using this approach, one

can use soft X-ray spectromicroscopy to study the role of bacteria and

their bioﬁ lms in changing the reduction/oxidation state or sequestra-

tion of various elements in the environment (Lawrence et al., 2003;

Yoon et al., 2004; Hitchcock et al., 2005) (see Figure 13–33), the growth

of crystaline materials (Chan etal., 2004), and other geochemical reac-

tions (Myneni et al., 1997; Tonner et al., 1999; Pecher et al., 2003). Spec-

tromicroscopy at the carbon edge can be used to study a variety of

organic processes, ranging from the diagenetic breakdown of organic

material over geological timescales and its presence and preservation

in fossilized plants and wood (Cody, 2000; Boyce et al., 2002, 2004) and

coals (Botto et al., 1994; Cody et al., 1995), and the role of natural

organic matter in the properties of soils (Thieme et al., 1994; Thieme

and Niemeyer, 1998a; Scheinost et al., 2001; Schäfer et al., 2003; Solomon

et al., 2005) including its role in the groundwater transport of radionu-

clides (Schäfer et al., 2005) (see Figure 13–34). Tomography has also

been used to study bacterial “microhabitats” (Thieme et al., 2003).

Other studies have considered the functional groups present in the

soot produced by combustion in diesel engines (Braun et al., 2004).

The trace element mapping capabilities of X-ray microprobes are also

very useful for studies in environmental science. Low concentration of

iron sets a biotic limit to carbon uptake in the southern Paciﬁ c; Twining

et al. (2003) have used microprobe studies to investigate this on a cell-

by-cell basis (see Figure 13–35) since bulk chemistry measurements do

5 µm

346.0 eV

weak Ca absoprtion

352.0 eV

strong Ca absoprtion

284.0 eV

weak C absoprtion

290.0 eV

strong C absoprtion

Figure 13–32. Images of a colloidal chemistry sample consisting of oil in water with clays and

calcium-rich layered double hydroxides used to “cage” the oil droplet where present (left and bottom

edges of the droplet). This illustrates the ability to highlight various elemental components in a room

temperature wet specimen. (Courtesy of Neuhäusler, 1999.)

Chapter 13 Principles and Applications of Zone Plate X-Ray Microscopes 897

5 µm

Protein

Polysaccharide

Lipids

Ener

(

)

290 300 310

Linear absorption

2 m

-1

protein

poly-

saccharide

lipid

silicate

CaCO

Figure 13–33. Quantitative chemical maps of protein, K

, lipids, and polysaccharides from a wet

microbial colony from the South Saskatchewan river, derived from STXM images (880 × 880 pixels)

and image sequences (52 energies, 230 × 230 pixels). The spatial distributions of the various chemical

species are determined by ﬁ tting the spectra from each pixel with a linear combination of the absorp-

tion spectra of the constituents. X-ray absorption spectra in the C 1 s region are shown for CaCO

, K

silicate, lipid, polysaccharides, and protein. The spectrum of CaCO

is from pure material. Those of

the other ﬁ ve species are derived from the C 1 s image sequence recorded from this bioﬁ lm using pixel

identiﬁ cation and (for lipid, polysaccharides) spectral subtractions based on ﬁ ts of the image sequence

to the spectra of pure reference materials. (Courtesy A.P. Hitchcock, McMaster University.)

not allow one to differentiate between protist types and particulate

matter at the same size scale. Other studies using zone plate micro-

probes have concentrated on the speciation of metals near the roots of

healthy and diseased plants (Yun et al., 1998b), the presence of metals

in soil bacteria (Kemner et al., 2004), sulfur speciation in bacteria

(Labrenz et al., 2000), in natural silicate glasses (Bonnin-Mosbah et al.,

2002), and in microbial ﬁ laments (Foriel et al., 2004), and elemental

concentrations in atmospheric particles (Ma et al., 2004). These repre-

sent only early examples, as the number of projects being carried out

using zone plate microprobes is increasing rapidly.

4.3 Materials Science

Applications of X-ray microscopes to material science include four

broad categories of study: chemical state mapping in polymer systems