¿ìè¶ÌÊÓƵ

For most people, the word ‘microscope’ probably conjures up an image
of an optical instrument through which you peer to see an enlarged specimen.
Yet the past 50 years have seen the development of several new kinds of
microscope that far exceed the limitations of conventional optical microscopes.
Everyone has seen the remarkable pictures of cells and other minute objects
taken with an electron microscope. The latest microscopes can focus on even
smaller objects, some resolving structures at the level of atoms. Others
can analyse specimens, producing images that reveal, for example, the detailed
distribution of particular chemical elements.

One of the most exciting analytical tools of this kind to have emerged
during the past 20 years is the ‘nuclear microscope’, which instead of relying
on light or electrons to scan a specimen, uses a well-focused beam of energetic
nuclear particles. These particles are the nuclei of atoms, and can be simple
protons (the nuclei of hydrogen), the nuclei of deuterium (‘heavy’ hydrogen
in which the nucleus contains a neutron in addition to the proton), or the
nuclei of helium or still heavier elements. The interaction of the beam
of particles with the specimen yields its own brand of information at a
microscopic level, and so makes the nuclear microscope a valuable addition
to the range of microscopes.

The nuclear microscope is still in its infancy compared with the established
electron microscope, and its underlying technology is rapidly evolving.
At present, there are between 30 and 40 facilities throughout the world,
all in various stages of development. The most common, and probably the
most well-developed type, is the proton microscope – it is already finding
important uses in medicine and electronics.

In a proton microscope, low energy protons are injected into a small
particle accelerator – usually a Van de Graaff machine which accelerates
the protons through electrostatic fields of several million volts. The energised
protons emerge from the accelerator in a beam several millimetres across.
To be useful for microscopy, this beam must then be focused down more than
a thousand times, to a diameter of a few micrometres or less. This finely
focused beam is then scanned across the specimen to yield detailed information
about its structure.

One of the major technical difficulties in building a proton microscope
lies in adequately focusing the beam. Protons are some 1800 times heavier
than electrons, so the technology developed for electron microscopes is
no help. However, various groups around the world have succeeded in building
appropriate magnets for squeezing down a proton beam, so that protons microscopes
have become a reality. Moreover, these instruments have shown that proton
microscopy provides information that would be impossible or very difficult
to acquire with other techniques. This justifies the continuing effort to
develop the technology, which is necessarily more complex than in other
types of microscope.

The first successful focused proton microscope was set up in 1969 by
John Cookson and Frank Pilling at the UK Atomic Energy Authority at Harwell.
They ultimately achieved a beam spot 3 micrometres across. Ten years later,
at the University of Oxford, we (Frank Watt and Geoff Grime) produced the
first beam to cross the sub-micrometre threshold, using special strong-focusing
magnets that are virtually free from aberration. Since then, the proton
microscope has come of age, its strength lying in microanalysis and its
ability to characterise materials. The Oxford Proton Scanning Microscope
(SPM) was the first to be set up solely for interdisciplinary applications,
with the aid of a grant from the Wellcome Trust to investigate neurodegenerative
diseases. Its current applications cover the gamut of major scientific disciplines,
from medicine to microelectronics, environmental studies to art history.

The proton microscope can provide unique information in three basic
ways. Energetic protons in a beam travelling through a material can collide
both with electrons and with nuclei in the surrounding atoms. The different
kinds of interaction yield different information about the sample, and together
provide a powerful means of microscopical analysis.

When a proton collides with an atomic electron it can knock the electron
right out of the atom, leaving a vacancy in the atomic orbitals. If the
electron is from an inner orbital, another electron from an outer orbital
will takes its place, but to do so it must lose energy which it does by
emitting an X-ray. Because the pattern of electron orbitals characterises
the atom of a specific element, so too does the energy of X-rays emitted
in this way. For example, a proton colliding with an iron atom produces
an X-ray with an energy of 6.4 kilo electronvolts (keV), while a collision
with a calcium atom emits an X-ray of 3.7 keV. Detecting the X-rays produced,
as a proton beam scans a specimen, maps the spread of chemical elements
in a sample.

This technique is known as proton induced X-ray emission, or PIXE, and
it began in 1969 with the work of Sven Johansson and colleagues at the Lund
Institute of Technology in Sweden. Since then, PIXE has developed into a
high precision technique, yielding maps of elements at resolutions of better
than a micrometre. At Oxford, for example, researchers have used the proton
microscope facility to study PIXE in a wide range of samples, from fly ash,
where the technique reveals the distribution of hazardous elements, to brain
tissue from people suffering from Alzheimer’s disease .

One of the advantages of PIXE is that it can routinely analyse the concentration
of elements to parts per million in many types of sample. This high sensitivity
is due to the low background of incidental X-rays that the protons produce,
which is 100 to 1000 times smaller than the background that arises when
electron beams are used to induce X-ray emission. An added attraction is
that PIXE involves only collisions with electrons in the inner core of the
atom, and so results do not depend on any chemical effects, which influence
only the outer electron shells. Analysis with PIXE is therefore quantitatively
accurate, with some researchers quoting accuracies as good as 2 per cent.

When energetic protons pass through thin samples, typically 50 micrometres
thick or less, they lose energy approximately in proportion to the number
of collisions they suffer with electrons (only a small fraction undergoes
collisions with atomic nuclei). This means that it is quite easy to calculate
the amount of energy that a proton of given energy loses as it travels through
materials of various densities. So, measurements of the energy lost as a
beam of protons passes through a specimen reveals the density at each point
as the beam is scanned across the sample. This technique of Scanning Transmission
Ion Microscopy, or STIM, allows contour maps of energy loss, or equivalently,
density, to be produced very quickly.

The advantage of STIM over similar techniques with other probes is due
to the relatively large mass of the proton – the very feature that makes
a proton microscope more difficult to build in the first place. Their greater
mass makes the protons more penetrating and means that they do not scatter
easily within a material, so that the beam tends to retain its hard-won
small size. For transmission electron microscopy, for example, specimens
must be as thin as 0.1 micrometres for the electrons to penetrate through
the sample and not spread out too far. Work at Oxford with STIM has shown
its benefits in studying semiconductor microcircuits, revealing features
that other techniques literally cannot reach .

A third method of using proton beams for microscopy involves the interactions
of the protons with the atomic nuclei in a material. Most of the proton
collisions are with electrons, but occasionally a proton will collide with
a nucleus and rebound, as if in a game of subatomic billiards. As Ernest
Rutherford discovered in the early part of the century, the energy with
which the proton rebounds depends on the mass of the nucleus it struck.
By measuring the energy of the rebounding, ‘back scattered’ proton, it is
relatively easy to determine the mass and, therefore, the type of nucleus
it hit.

Mapping the light elements

Rutherford back scattering spectrometry, or RBS, as this technique is
known, provides an alternative, independent method of analysing elements
in a sample. In practice, RBS complements PIXE beautifully, for whereas
PIXE works best with elements in the region of sodium or heavier, RBS can
resolve and analyse lighter elements in thin specimens.

But RBS does more than analyse the lighter elements. When a proton collides
with a nucleus, this nucleus may be buried deep within the sample. The scattered
proton, therefore, loses some energy as it makes its way back out of the
sample. However, as with STIM, this energy loss is easy to calculate, so
by studying how the energies of the back scattered protons have shifted
to values slightly below those characteristic of collisions with nuclei
of specific elements, you can find out how deep the struck nucleus is within
the sample. In this way, RBS provides a three-dimensional, non-destructive
means of mapping elements. Moreover, as with PIXE, RBS is not influenced
by the chemical state of the sample.

Nuclear microscopy, and in particular the proton microscope, is probably
at about the same point in its development as the electron microscope was
in the early 1950s. Instruments are not yet available commercially in a
complete ‘user friendly’ package, and most facilities that exist today are
located in physics laboratories. It is important that we should still regard
the ‘state of the art’ proton microscope as a development in applied physics,
albeit with the potential to solve many problems in a wide variety of scientific
disciplines. There are still many potentially exciting developments to explore,
from improving the spatial resolution of the microscope so that it can reveal
details of atomic dimensions, to bringing a highly focused proton beam into
air, outside the envelope that keeps it in a vacuum. The proton microscope
has an exciting future ahead.

Frank Watt and Geoff Grime initiated and developed the Scanning Proton
Microscope Unit at the University of Oxford.

* * *

1: The proton microscope and Alzheimer’s disease

Alzheimer’s disease is a neurodegenerative disease that affects a significant
number of people in their old age. In recent years, a controversy has arisen
over the role, if any, of aluminium in the disease (¿ìè¶ÌÊÓƵ, 21 July
1990, p 21). Realising that proton microscopy could provide definitive analytical
evidence, since 1987 the Wellcome Trust has funded a programme of research
with the Oxford Scanning Proton Microscope Unit, which is aimed at clarifying
the issue. Judith Landsberg, a member of the SPM team, is taking part in
this work together with Brendan McDonald and Margaret Esiri from the Department
of Neuropathology in Oxford.

One problem that hinders work in this field is that aluminium, in the
form of aluminosilicates, is an extremely common form of contamination,
much of it originating in soil dust borne by the air. In conventional microanalysis,
the pathological features of Alzheimer’s disease in brain tissues (senile
plaques and neurofibrillary tangles) must be identified by chemical staining
before analysis can take place. The staining procedure may cause contamination,
or redistribute elements within the specimen.

The team at Oxford has taken the view that if they can identify the
pathological features without using chemical treatment of any kind, they
stand a better chance of producing definitive results in any subsequent
microanalysis. It turns out that the complementary techniques of proton
microscopy make it ideally suited to such an approach.

Opposite is a photomicrograph (taken with an optical microscope) of
a section of tissue showing a senile plaque, a major feature of Alzheimer’s
disease. The tissue has been treated with an antibody stain to reveal the
plaque, which shows the classic structure of a central core surrounded by
a fuzzy halo.

Figure 1a shows a STIM image, taken by scanning transmission ion microscopy
(STIM)-that is a density contour map produced by the SPM, of a similar structure,
this time in untreated tissue. It reveals a dense central core with a less
dense outer halo. The SPM can simultaneously produce PIXE images, which
map the distributions of sodium and heavier elements, such as phosphorus
and sulphur, as shown in Figures 1b and 1c. Because the samples the team
analyses are thin, they can also use the technique of RBS to give maps of
lighter elements, specifically carbon and nitrogen (Figures 1d and 1e).

Using proton microscopy, the team can seek out and identify specific
features in unstained, untreated tissue, and simultaneously map all elements
to high levels of sensitivity and accuracy. With this powerful technique,
they hope particularly to resolve the issue of the role of aluminium in
Alzheimer’s disease, but also to solve other problems in medicine and biology.

* * *

2: Proton microscopes and microelectronics

Mark Breese, a research fellow at Oxford, and Phil King, a postgraduate
student, have been using the SPM to study defects in semiconductor microcircuits.
In the example shown here, they found that the complementary techniques
available with proton microscopy allowed them to identify a flaw in a complicated
circuit, which other techniques had failed to reveal.

A microcircuit is a three-dimensional structure with various metallic
and insulating layers. Figure 2 shows the way in which the different metallic
layers should have been laid down as the circuit was constructed. In addition,
the surface of the circuit (with the exception of the contact pads at the
right on the figure) is covered with a protective layer of silicon dioxide.
This layer is between 1 and 2 micrometres thick, which means that it is
difficult to reach the circuit beneath with an electron probe to analyse
it any detail, without etching the silicon dioxide away. Electrons can penetrate
only between 1 and 2 micrometres into solid materials.

With the proton microscope, the silicon-dioxide is no obstacle. Figure
3 shows that by using STIM, Breese and King could map the smallest structures
in the microcircuit. These maps show the average energy loss of protons
passing through the circuit, with the dark regions representing high energy
losses and the light regions low energy loss. The thick tracks of tungsten
produce the greatest energy loss and, therefore, show up well on these STIM
images.

Breese and King can also choose to map a specific range in energy loss,
as opposed to the average energy loss, to reveal particular features of
the microcircuit. For example, Figure 4a is a map of low proton energy loss,
and it shows evidence of a misplaced track. Comparing this image with Figure
2 reveals that one of the tracks is slightly offset from the centre of the
gap in the layer of tungsten and titanium.

In this way, STIM revealed a likely defect, but it was possible for
the researchers to go further. They could use PIXE to determine the nature
of the fault that appears in the STIM image. Figures 4b and 4c show maps
of tungsten and titanium in the same region. They indicate that the fault
is in the positioning of the 1-micrometre wide track of titanium.

In addition, the researchers can bring RBS, the third of the three techniques
available with the SPM, to bear upon the circuit. This allows them to measure
simultaneously the thicknesses of the various layers of metallisation. In
this sample, they found that the aluminium layer was 0.9 micrometres thick,
the tungsten layer 0.7 micrometres thick, while the silicon-dioxide surface
layer was 1.1 micrometres thick. This shows how the three techniques together
allowed the researchers to carry out ‘reverse engineering’ – to discover
how the circuit was built without physically dismantling it, a valuable
procedure for detecting defects.