Blake A.J.(ed.) Crystal Structure Analysis

Подождите немного. Документ загружается.

This page intentionally left blank

X-ray and neutron

sources

William Clegg

22.1 Introduction

The main topic of this book is the analysis of crystal structures using

diffraction of X-rays by single crystals. Most such experiments are

carried out in local research laboratories, either in Universities or in

industry, and make use of commercial X-ray diffractometers that are

equipped with ‘X-ray tubes’ of various designs. These involve the gen-

eration of X-rays by directing fast-moving beams of electrons at metal

targets, and typically consume electrical power in the range from tens

of watts to tens of kilowatts.

Much higher X-ray intensities are available from large-scale national

and international storage-ring facilities (more commonly, if strictly

incorrectly, known as synchrotron facilities), and there are ways other

than the enhanced intensityin which the properties of theseX-rays differ

from those from laboratory X-ray tubes.

Furthermore, diffraction from crystals can occur with fast-moving

beamsofsubatomicparticles,particularly neutrons and electrons,which

have wave properties and interact with the investigated crystalline

material on quite different physical bases fromX-rays, giving diffraction

patterns with different information content.

In this chapter we consider the conventional and synchrotron sources

of X-rays and their different properties and uses, then we discuss the

application of neutron diffraction.Electrondiffraction by solids, because

it involves a much stronger interaction with the sample, is normally

conﬁned to thin specimens and surfaces, and it ﬁnds wide application

in electron microscopy. It, and the use of electrons for diffraction by gas

samples, lie outside the scope of this text, and will not be considered.

22.2 Laboratory X-ray sources

Most laboratory X-ray diffractometers, whether ﬁtted with serial detec-

tors or area detectors, operate with a conventional sealed X-ray tube, the

333

334 X-ray and neutron sources

basic design of which has not changed for a long time, though ceramic

insulating materials are increasingly being used instead of glass. The

principle of operation of an X-ray tube is simple (Fig. 22.1). Electrons

are generated in a vacuum by passing an electric current through a wire

ﬁlament and are accelerated to a high velocity by an electric poten-

tial of tens of thousands of volts across a space of a few millimetres.

The ﬁlament is held at a large negative potential and the electrons are

attracted to an earthed and water-cooled metal block, where they are

brought to an abrupt halt. Most of the kinetic energy of the electrons

is converted to heat and carried away in the cooling water, but a small

proportion generates X-rays by interaction with the atoms in the metal

target. Some of the interactions produce a broadrange of wavelengths of

X-rays, with a minimum wavelength (maximum photon energy) set by

the kinetic energy of the electrons. For our purposes, however, the most

important process is the ionization of an electron from a core orbital,

followed by relaxation of an electron from a higher orbital to ﬁll the

vacancy. This electron transition leads to loss of excess energy by radi-

ation, and the emitted radiation, with a deﬁnite wavelength, is in the

X-ray region of the spectrum. Several different transitions are possi-

ble, so a number of intense sharp maxima (in wavelength terms) are

superimposed on the broad output of the overall spectrum of radiation

produced (Fig. 22.2).

Cooling water

X-rays

Electrons

Fig. 22.1 Schematic diagram of a sealed

X-ray tube.

Intensity

Fig. 22.2 The spectrum of X-rays gener-

ated by a sealed tube.

Since we use monochromatic X-rays for most purposes, one particular

intense line of the output spectrum is selected and the rest discarded.

The usual way of achieving this is to use diffraction itself: the X-rays

emerging from a thin beryllium window in the X-ray tube are passed

through a single crystal of a strongly diffracting material set at an appro-

priate angle. The (002) reﬂection of graphite is very widely used, with

a2θ angle of just over 12

◦

for Mo-Kα radiation (λ = 0.71073 Å, the

most intense line from a target consisting of molybdenum metal) and

about 26.5

◦

for Cu-Kα radiation (λ = 1.54184 Å, from a copper target).

All other wavelengths pass undeﬂected through the monochromator

crystal, leaving a single wavelength for the diffraction experiment.

Various developments of the basic X-ray tube produce higher inten-

sities. The main limitation is the amount of heat produced, which can

damage or even melt the target if it is excessive. One way to reduce the

heat loading, and hence allow larger electron beam currents and more

intense X-rays, is to keep the target moving in its own plane by rotating

it, so that the target spot is constantly replaced. Rotating-anode sources

require continuous evacuation because of the moving parts, and their

use involves much more maintenance as well as higher energy con-

sumption than sealed tubes. The increase in intensity can be up to about

a factor of ten.

Another approach is to collect and concentrate more of the X-rays

generated instead of just the narrow beam taken from a standard tube.

Recent technological advances have produced extremely well polished

mirrors giving glancing-angle total reﬂection of X-rays, and devices

consisting of variable-thickness layers of materials with different crystal

22.3 Synchrotron X-ray sources 335

latticespacingstogivea focusing effect throughdiffraction(also referred

to, not strictly correctly, as mirrors). X-rays can also be concentrated

and focused through glass capillaries. Each of these devices can give

a signiﬁcant increase in intensity from a suitable X-ray tube (whether

sealed or rotating anode). Some of them can be combined with X-ray

tubes in which the electron beam is magnetically focused to give a very

small target spot, reducing heat loading, and these are known as micro-

focus tubes. They typically operate with power consumption of tens of

watts, compared with 1–3 kW for conventional sealed tubes and 5–30

kW for rotating-anode sources. Overall, the developments in X-ray gen-

eration and focusing have led to increases of 1–2 orders of magnitude

for laboratory X-ray sources.

22.3 Synchrotron X-ray sources

When moving charged particles are deﬂected by a magnetic ﬁeld, they

emit electromagnetic radiation and hence lose energy. The radiation

wavelength depends on the particles (charge, mass and velocity) and

on the magnetic ﬁeld strength. This effect was found to occur in syn-

chrotrons, particle accelerators with a closed path (pseudo-circular,

actually polygonal resulting from an array of magnets), and is an unde-

sirable feature in the orginal primary purpose of such devices. However,

the radiation has special properties and can be exploited for diffraction,

spectroscopy and other uses. The ﬁrst such experiments were carried

out around 1970 on the NINA accelerator at Daresbury, UK, and gave

rise to the term ‘ﬁrst-generation’ synchrotron radiation source, referring

to parasitic use of a synchrotron that was designed for other purposes.

Such a radiation source is far from ideal, being erratic and unstable

through rapid acceleration, deceleration and deliberate collisions of the

particles; a storage ring, with a stable circulating particle beam, is much

better suited as a reliable source of electromagnetic radiation. Once the

usefulness of synchrotron radiation (SR) had been established, storage

rings were designed and built, dedicated to the production of SR, with

the aim of stability in terms of intensity, position and direction of the

radiated beams. These are called second-generation SR sources, and ini-

tially the SR was generated exclusively by the bending magnets that

keep the particles in their circulating motion. The energy of the parti-

cles is maintained by the inclusion of microwave cavities at one or more

positions in the ring, compensating for the energy loss associated with

SR emission. The ﬁrst second-generation SR source was the UK Syn-

chrotron Radiation Source (SRS), built at Daresbury to replace NINA,

and operational from 1980 to 2008.

Even greater intensity (and some other useful properties) can be

achieved by putting more complex arrays of magnets in the straight

sections between bending magnets. These ‘insertion devices’ include

wigglers and undulators, which induce a series of sideways oscillations

in the particle beam path; each wiggle generates SR and the individual

336 X-ray and neutron sources

Fig. 22.3 Diamond Light Source. Copyright Diamond Light Source, reproduced with

permission.

contributions combine in different ways depending on the precise

arrangements of the magnets. Some insertion devices were added to

second-generation sources such as SRS, but they provide the main SR

output of third-generation sources, the bending magnets of which can

also be used, of course. Most SR sources now operational are of this

third generation, including Diamond (Fig. 22.3), the UK replacement

for SRS, many other national facilities, and international sources such as

the European Synchrotron Radiation Source(ESRF) in Grenoble, France.

The fourth generation of SR sources, currently under development, is

based on the free-electron laser.

The precise properties of SR depend on a combination of factors,

including

(a) the energy of the stored particle beam (several GeV, with essen-

tially the speed of light, resulting in relativistic behaviour);

(b) the size of the storage ring, the number of bending magnets, and

details of other magnetic components for controlling and focusing

the particle beam;

(d) the structure of the particle beam, which usually consists of dis-

crete bunches rather than a continuous stream, giving rise to a

rapid pulse behaviour of the SR output;

(e) the way in which beam-current loss (inevitable through collisions,

imperfect vacuum and other effects) is dealt with, in various ‘top-

up’ and reﬁll operations;

(f) optical components for conditioning the SR, including monochro-

mators and mirrors for wavelength selection, harmonic rejection

and focusing.

22.3 Synchrotron X-ray sources 337

SR is produced tangentially every time the particle beam changes

direction in bending magnets or insertion devices (Fig. 22.4). As a result

of relativity, it is strongly concentrated in a single forward direction, giv-

ing a very highly collimated beam of radiation. It is almost completely

polarized in the plane of the storage ring, has a very high intensity

compared with conventional X-ray sources, and covers a continuous

wide spectrum from infrared to hard X-rays (undulators give a jagged

stepped X-ray spectrum), with a maximum photon energy (minimum

wavelength) dictated by the operating conditions. For the purposes of

X-ray crystallography, we can think of a synchrotron storage ring in sim-

ple terms as a large device that exploits relativity to convert microwave

energy into X-rays by a massive doppler shift. Any wavelength can

be selected from the broad spectrum by a monochromator, or the con-

tinuous ‘white’ X-ray spectrum can be used for the Laue diffraction

technique, which is not discussed in this book.

Electron beam

Magnets

nchrotron radiation

Fig. 22.4 The principle of operation of a

synchrotron storage ring.

The very high intensity, several orders of magnitude greater than

from conventional sources, is the most obvious and desirable feature

of SR X-rays. It allows diffraction patterns to be measured quickly, even

from tiny crystals (down to micrometre dimensions, depending on the

chemical composition and crystal quality of the sample), or from other

samples giving relatively weak diffraction as a result of structural faults

such as disorder. Obviously this is useful when larger single crystals

can not be obtained, and individual powder grains can be treated as

single crystals, though therequirements of crystalmounting and diffrac-

tometer mechanical precision are demanding; it is also an advantage

for chemically unstable and sensitive materials, and makes it possible

(in combination with the pulsed nature of SR and the use of lasers) to

investigate short-lived excited states.

The advantages of the high intensity of SR are further enhanced

by the high degree of collimation, which makes individual reﬂections

from single-crystal samples stand out moreclearly from the background

because of their sharper proﬁles; these are dictated largely by the sam-

ple quality rather than a non-parallel incident X-ray beam. This can

also help in the spatial resolution of reﬂections from a sample with a

large unit cell, or if a short wavelength is selected. A short wavelength

gives access to higher-resolution data for charge-density studies, can

reduce some systematic errors such as absorption and extinction, and

can allow more of the diffraction pattern to be measured from a sample

in a diamond anvil high-pressurecell or other special environment. Con-

versely, a longer wavelength spreads out a dense diffraction pattern for

a large structure. A particular wavelength might also be chosen to min-

imize or to maximize special effects such as anomalous scattering. Bent

monochromators and mirrors (giving total external reﬂection of X-rays

at a glancing angle) can be used to focus and concentrate the available

X-rays to match the sample size, so that the high ﬂux is more effectively

usedashigh brilliance (ﬂux in a given cross-sectionalareaor solid angle).

The pulsed nature of SR is exploited in special time-resolved stud-

ies, but is not relevant to most crystallographic users. The polarization

338 X-ray and neutron sources

properties mean that diffractometers have to be operated ‘on their side’,

with diffraction measured in a vertical rather than a horizontal plane,

making synchrotron installations look rather strange compared with

standard laboratory setups.

SR sources are large national and international facilities providing

equipment and support for a wide range of scattering, spectroscopy,

imaging and other applications. Modes of access vary, but usually

include some form of peer-reviewed application process on a regular

basis (often twice per year), leading to use that is ‘free at the point of

access’ to successful academic research groups, and charged for com-

mercial users. There may be service modes of operation, or users may

have to carry out their own experiments after appropriate training for

safety and other aspects.

Structure determination with SR single-crystal diffraction facilities

has been particularly important in certain research areas where small

crystals and other weakly scattering samples frequently occur. These

include microporous materials, often synthesized solvothermally; poly-

meric co-ordination networks; other supramolecular assemblies with

weakintermolecular interactions; pharmaceuticals; pigments; low-yield

products of reactions; and structures with several molecules in the

asymmetric unit (Z



> 1).

Aparticular problemfrequentlyencounteredis of crystals that growto

a reasonable size in one or two dimensions, but form only very thin nee-

dles or plates with a total volume, and hence scattering power, below

acceptable levels for laboratory study. In addition, there are applica-

tions where it is an advantage to select as small a crystal as possible to

avoid other problems (e.g. in high-pressure studies, or to reduce absorp-

tion and extinction effects). The use of synchrotron radiation opens up

the possibility of determining a complete structure from a single pow-

der grain and thus investigating the homogeneity of a microcrystalline

sample.

Some materials form only very small crystals because of poor crys-

tallinity, but even large crystals may be of inferior quality, with a large

mosaic spread, and so give broad and weak reﬂections. Synchrotron

radiation can give adequate diffracted intensities, and the low intrin-

sic beam divergence also minimizes the breadth of observed reﬂections.

Weak diffractionmay be caused particularly by various types of disorder

in the structure, and cooling of the sample is alsoimportant.Asomewhat

related topic is the study of substructure/superstructure relationships.

Resolution of such structures depends critically on the measurement

of very weak reﬂections that alone distinguish different possible space

groups.

For unstable species and time-resolved studies, very high inten-

sity means that data can be collected at maximum speed while still

achieving an acceptable precision of measurement. The combination of

synchrotron radiation and high-speed area detectors provides a means

of doing this type of experiment. Rapid data collection is essential for

unstable samples, but can also be very useful in collecting multiple data

22.4 Neutron sources 339

sets for a sample under different conditions of temperature or pressure

in order to investigate the effects of varying these conditions. It may be

possible to follow solid-state reactions, where the reactant and product

have related structures and crystal integrity is maintained in the reac-

tion; these may include phase transitions, polymerization, and reactions

related to catalysis.

Although SR facilities are expensive to build and operate, their use in

structure determination can be very cost effective, with rapid through-

put of samples, data and results. Typical use of station 9.8 at Daresbury

SRS (ﬁnally closed in August 2008 after about 12 years of operation) by

the UK National Crystallography Service has given around 12–15 full

data sets in each 24-hour period, investigating samples that had been

previously screened and found to be beyond the capabilities of even the

most powerful conventional laboratory sources, and results have been

published in leading international journals. Even higher productivity is

expected at Diamond beamline I19.

For a historical survey of SR work, see Helliwell (1998). For a more

extensive account of SR in crystal structure determination, see Clegg

(2000), and for information on the use of SR in the UK National

Crystallography Service, see http://www.ncl.ac.uk/xraycry.

22.4 Neutron sources

X-rays are used for crystal-structure determination because they have

a wavelength comparable to the size of molecules and their separation

in solids, so they give measurable diffraction effects from crystals. By

deﬁnition, they are the only region of the electromagnetic spectrum with

an appropriate wavelength. However, a beam of neutrons can have a

similar associated wavelength, related to its velocity v and momentum

mv (where m is the neutron mass) by λ = h/mv. Such neutrons may

be generated by nuclear reactors and by spallation sources, which are

described later. Such large-scale facilities are, of course, an expensive

way to produce neutron radiation, so it is worthwhile only if there are

clear advantages over X-rays. In the case of scattering and diffraction

this is true in certain cases, and the two techniques are complementary.

X-rays are scattered by the electron density of atoms, so the scatter-

ing is proportional to atomic number. This means that ‘heavy atoms’

(those with many electrons) dominate X-ray diffraction by crystals, and

lighter atoms are relatively difﬁcult to see and imprecisely located. It

also means that neighbouring elements in the periodic table give almost

identical X-ray scattering and can not easily be distinguished on this

basis alone.

Neutrons, by contrast, are scattered by atomic nuclei. There is no

simple dependence on atomic number, and the variation of neutron

scattering across the whole periodic table is much smaller than that of

X-ray scattering. Neutron scattering by neighbouring elements can be

very different; the variation from element to element is quite erratic, and

340 X-ray and neutron sources

differentisotopesofagiven element usually have quite differentneutron

scattering powers, while they are completely indistinguishable to X-

rays. These differences are illustrated by some examples in Table 22.1.

Note that some nuclei have a negative neutron scattering factor (usually

expressed as a scattering length with units of fm); they scatter exactly

out of phase with other nuclei. In general, neutron scattering is much

weaker than X-ray diffraction (the two sets of values in Table 22.1 are not

on the same scale). The adjacent elements Co and Fe have very different

neutron scattering factors, while their X-ray scattering factors differ by

only a few per cent. Deuterium scatters neutrons almost as strongly as

does uranium, but is essentially invisible to X-rays when both elements

are in the same material, and it is dramatically different from its lighter

isotope H for neutron diffraction.

Table 22.1. Selected X-ray

and neutron scattering

factors (electrons and fm,

respectively)

Element X-ray Neutron

H1−3.74

D 1 6.67

O 8 5.81

V23−0.38

Fe 26 9.45

Co 27 2.78

Ba 56 5.28

U 92 8.42

These features of neutron scattering by nuclei lead to a number of

practical advantages over X-ray diffraction in certain types of studies.

(a) It is often easier to locate light atoms precisely in the pres-

ence of heavier ones. For example, the exact positions of oxide

anions in complexmetal-oxide structurescan be veryimportant in

understanding properties such as superconductivity and unusual

magnetism. If heavy metals are present, this is a serious challenge

for X-ray diffraction, especially if the oxide positions display any

disorder. Oxygen has a relatively large neutron scattering length;

it is very similar to that of barium (Table 22.1) – in fact a little

higher – but with X-rays Ba scatters seven times as strongly as O.

(b) The most extreme case of this is the location of H atoms, for which

X-ray diffraction is not the ideal technique in view of the low elec-

tron density of H. Neutron diffraction is very much more effective

here(evenmoresoif DreplacesH),incases whereit isimportantto

ﬁnd H atoms reliably, such as in metal-hydride complexes, agostic

interactions, and unusual hydrogen-bonding patterns. The prob-

lem for X-ray diffractionis madeworse by the fact that the electron

density of the H atom is involved in bonding; it is not centred on

the nucleus, but is distorted towards the adjacent atom, leading to

a systematic apparent shortening of X–H bonds in X-ray diffrac-

tion studies. Neutron diffraction is not affected by the bonding or

by other valence-electron density features such as lone pairs, and

it determines accurate nuclear positions and hence internuclear

distances.

atoms of neighbouring elements in the periodic table is impor-

tant for a reliable structure determination, such as in mixed-metal

complexes or metal alloys.

(d) Isotopes of the same element appear quite different in neutron

diffraction,in most cases, whereas they are completely identical in

X-ray diffraction. This may be useful, for example, in establishing

the positions of isotopically labelled atoms in a product as part of

an investigation of the reaction mechanism.

22.4 Neutron sources 341

The weak interaction of neutrons with nuclei in a crystalline sample,

compared with the rather stronger interaction of X-rays with electrons,

means that larger crystals are usually required for neutron diffraction,

though this limitation of the technique is mitigated to some extent with

more intense modern neutron sources and more sensitive detectors. On

the other hand, it means that neutrons are generally more penetrating

than X-rays, giving lower absorption, allowing the study of materials

in containers through which X-rays would not pass, and leading to lit-

tle radiation damage (except in cases where neutrons react with some

nuclei to generate new nuclei). Nuclei are also effectively point scatter-

ers in contrast to the ﬁnite size of the electron distribution in an atom, so

neutron scattering from a stationary atom does not fall off with increas-

ing angle as does X-ray diffraction – the θ dependence is due only to

atomic displacements, which are reduced at low temperature.

One other special property of neutrons is that they have an instrin-

sic magnetic moment, or spin. This can be exploited to investigate

magnetic properties such as ferromagnetism, ferrimagnetism and anti-

ferromagnetism, which involve regular arrangements of atomic mag-

netic moments (and these are due to unpaired electrons, so this is a

neutron-electron interaction) in a solid material.

Two main types of neutron sources are used for crystallography. A

nuclear reactor uses neutrons to maintain its activity, but more are pro-

duced by

235

U ﬁssion than are needed for the continued nuclear chain

reaction, so the excess can be extracted in a continuous supply. The

Institut Laue-Langevin (ILL) in Grenoble is an example, and serves as a

European international facility, adjacent to ESRF.

Aspallation source generates rapid pulses of neutrons (and other sub-

atomic particles) by accelerating protons in a synchrotron and directing

them at a target containing heavy-metal atoms such as tungsten. The

wavelength of each neutron can be determined by measuring its ‘time

of ﬂight’ between the source and detector, as an alternative to select-

ing a monochromatic beam. An example of a spallation source is ISIS

at the Rutherford Appleton Laboratory in Oxfordshire, UK, adjacent to

Diamond.

For both types of neutron source, diffraction equipment is similar to

that used with X-rays, but it tends to be larger and more heavily shielded

for radiation. The greater penetrating power of neutrons means that

samples can be held in larger and thicker containers, including closed

low-temperature and high-pressure devices.

Finally, we note that combining X-ray and neutron diffraction for the

same sample material can exploit the complementary advantages of the

two techniques, for example in obtaining reliable structural information

when a wide range of elements is present, simultaneously investigating

geometricaland magnetic structural features,or decoupling valence and

atomic displacement effects in accurate high-resolution charge-density

studies of bonding.

For more detailed accounts of neutron diffraction, see Wilson (2000);

Piccoli et al. (2007).