In imaging using X-rays and neutrons, an object is placed between a source that emits radiation and a detector. The object itself partially attenuates the radiation by absorption or scattering. What the detector records depends on different factors including the type of radiation, the energy and current of the incident beam, and the properties and thickness of the material.
The nature of the X-ray interaction is with the atomic shell of the irradiated elements. One can imagine in a simplified way that when X-rays hit an atom, the energy that is absorbed lifts single electrons onto higher energetic levels or even causes one electron to leave the shell completely (ionization). Materials such as iron, copper, tin, lead, etc. consist of large atoms with a high number of electrons and high electron density. For higher X-ray energies, Compton scattering also plays an important role, meaning that not only the amount of electrons but also the material’s electron density are important. Dense materi-als can attenuate X-rays better than materimateri-als consisting of smaller atoms, such as carbon or hydrogen, which are the main elements of all organic components. How strongly a material attenuates radiation is represented by the specific attenuation coefficient (μ). The transmission of X-rays through a material is expressed by the Lambert-Beer law:
𝛪 = 𝛪 𝟢𝑒⁻�� (1)
In this equation 𝛪 𝟢 and 𝛪 are the beam intensities before and after transmission of the object, 𝑑 is the transmitted length of the material, and μ is the attenuation coefficient.
The neutron beam is produced either by fission in a nuclear reac-tor or by spallation, a process in which a highly-accelerated particle (e.g., proton) beam hits a heavy metal target from which neutrons are emitted due to the excited nuclei. Unlike X-rays, when a neutron beam hits the investigated material it interacts with the nucleus. The Lambert-Beer-law is also valid for the transmission of material. While 1. X-Ray and Neutron Imaging for Musical Instrument – Brief History
When Conrad Wilhelm Röntgen undertook his first experiments with X-rays in 1895, he noted the different absorption properties of the various materials. Organic materials like muscles were easily penetrat-ed by the radiation, while more dense materials like bones and metal proved to be strong absorbers. On paintings, areas covered with pig-ments containing lead were especially hard to penetrate by the newly discovered X-rays. The potential for this technology to revolutionize the examination of artwork was recognized from the very beginning, resulting in the first patent for the investigation of paintings in 1914 by Alexander Faber [1]. Only several decades later the technology was applied to the examination of musical instruments, beginning in 1949 with early X-ray images of a wooden oboe published by Halfpenny [2], but it has been used systematically ever since for the study of such instruments [3]. Shortly after the invention of computed tomography by Godfrey Hounsfield at the end of the 1960s, this technology that generates virtual sections through an object using X-ray transmission images from several angles was first applied for the examination of art-works in 1978 [4]. In the late 1980s, the potential of this technique was transferred to visualize the inner structure of musical instruments [5].
From 2014 to 2018, a team of scientists and conservators of the Ger-manisches Nationalmuseum in Nuremberg and the Fraunhofer EZRT in Fürth scanned more than 100 instruments during the MUSICES-project and published recommendations for conducting 3D CT scans.
Nowadays, X-ray computed tomography is an established tool for the examination of all kinds of cultural heritage objects.
In 1932, when X-rays were already a widespread tool for diagnostics in medical and (art) technological applications, an important discovery concerning the atomic model was made. Sir James Chadwick proved that the neutron exists as an elementary particle, providing the foundation for completing the explanation for the composition of the atomic nucleus.
Only three years later, the first successful experiments in neutron ra-diography were carried out by Hartmut Kallmann and Ernst Kuhn. In 1946, Otto Peter published the first images of industrial, mostly metallic objects. He observed that—unlike X-rays—organic materials and water are not easily penetrated by neutrons but cause a strong attenuation [6].
It took until the 1990s for the first results of neutron tomography to be published [7]. Until now, compared to X-ray imaging, only a small frac-tion of investigafrac-tions of cultural heritage objects take advantage of the properties of neutrons that mean metal objects can be penetrated more
testing. The spatial resolution can be very high. In the case of musi-cal instruments, a spatial resolution of 100 μm is usually sufficient. In the case of smaller objects or by using special detection methods, the resolution can also be better. Neutron CT is also used for industrial applications, although it is not so easily accessible, as such services are only available at a handful of research institutions with neutron sources and dedicated neutron imaging facilities. The resolution is limited to the order of tens of microns due to physical reasons, which is still enough for the examination of musical instruments.
2.3 Case Study 1: Scanning a Wooden Stringed Instrument
In order to compare the capacities of neutron and X-ray imaging for the examination of wooden instruments, a scan of a pardessus de viol made by Michel Colichon (Germanisches Nationalmuseum, inv. no. MIR782) was undertaken at the Paul Scherrer Institute (Villigen, Switzerland) using the thermal neutron imaging facility NEUTRA. The instrument was first scanned with neutron CT and then with X-ray CT at the exact same position. There are several interesting features in Colichon’s in-strument that can be the reason for such an investigation, among them the fact that the belly is not carved but assembled out of several pieces of bent wood. In this case, the instrument shall represent a number of possible interrogations that can concern wooden instruments, from violins and guitars to recorders or flutes, etc. In all of these cases, the investigator might want to have detailed information about the struc-ture of the instrument, the joints, the manufacturing process, the con-servational state or dimensions, e.g. in order to measure the thickness of a top plate or the diameter of a bore.
When comparing the results of both imaging methods the differ-ent qualities of transmission in the wood are quite obvious.
In a cross sections of the lower block [Fig.1], the X-ray image shows the exact orientation of the annual rings, which are (like in many instru-ments before 1800) parallel rather than vertical in relation to the back plate. It is also visible that the block has been assembled from two pieces with different orientations of the annual rings. The neutron image shows a brighter region at the edge of the material especially at thicker parts, the greyscale at the internal parts is much lower and no structure is visible.
Fig.1 Cross section of the lower block (left: neutron CT; right: X-ray CT).
Fig.1
beam, it is easy to penetrate heavy elements such as lead, whereas hydrogenous material such as wood and water are highly attenuating, showing the complementarity in comparison to X-rays.
Since both imaging techniques have to interact with matter, call-ing them “non-destructive” is only partly true. There are some cases in which an examination with one of the techniques can cause changes in the material. In the case of X-rays, the ionization effect can cause material changes [11]. For example, gemstones and glass can change colour under certain conditions. Also, information for thermolumines-cense dating of pottery can be deleted [12]. When neutrons are inter-acting with the atomic nuclei, it happens that the nucleus captures the neutron. This changes the isotopic status which can mean a radioactive activation of some materials. After exposing the object to the beam, the isotopes degrade again depending on their half-time-period. For cultural heritage objects, this can mean that the object has to be stored in a ray-proof environment for a while. Activation can affect several materials used in musical instruments, such as silver, gold and also the tin found in bronze. It is worth mentioning that both effects—ionization and activation—can also be used as detection methods for trace elements, e.g., X-ray fluorescence (XRF) or neutron activation analysis (NAA).
2.2 Computed Tomography – Essentials
Both imaging techniques fulfil the requirements concerning the interac-tion between beam and sample for use in computed tomography (CT).
This imaging method records a high number of single 2D radiographs at different angles which are used to compute tomograms, i.e. slice images, containing in depth and thus 3D information of the structure and content of an object. This 3D volume allows one to look inside the bodies of instruments and into the inner structure of the material.
Construction details and tool marks can be detected, cracks and other damages at hidden parts can be visualised, and measurements can be taken at any (otherwise inaccessible) location with high precision. The 3D data set can even be used for replication via 3D printing or CNC-milling. 3D-CT is currently the most powerful method for the structural examination of musical instruments. Here, only industrial and non-medical computed tomography is considered. While in non-medical X-ray tomography, the patient lies on a cot and the detector and source are rotating around him, the industrial setting is different. The object is placed on a rotation stage between the source and the detector. For X-ray CT, the detector is in most cases a flat panel with a fixed size, such as 40 cm x 40 cm. The setting can be adjusted to the size of the object and higher spatial resolutions are possible compared to medical CT.
X-ray computed tomography is widely used in the industrial sec-tor for monisec-toring production processes by means of non-destructive
(Fig.4), whereas the X-ray image only shows a more or less uniform surface. The contrast of the annual rings is also much higher, allowing for precise analysis using techniques such as dendrochronological dating.
2.4 Case Study 2: Scanning Brass Instruments
A 3D-CT scan of a brass instrument can provide important information not only about the construction, but also measurements like the exact total length of the bore also in case of difficult geometries or the inner volume. The shape of the bore and bell can be analysed to calculate the sound quality (brassiness). Scanning a brass instrument can be challeng-ing; the material’s density is high and the construction is not simple and homogenous. Some parts, such as the walls, are quite thin while other parts, such as the valves, are very thick.
As a representative for brass instruments, a mouthpiece was scanned with neutron tomography. The scan is compared to a scan of a mouthpiece which was executed during the MUSICES-project using X-ray computed tomography. A mouthpiece is an important part of a brass instrument because it can influence the sound of an instru-ment and many players have their favourite examples. With traditional methods, it is difficult to measure the exact shape of the bore, e.g.
for the purpose of a reproduction. In this case, the CT-data shall be examined considering three factors: first, how much information the images provide about the inner metal structure of the object; second, how exact measurements can be taken; and third, if the scan is suit-able for an exact reproduction of the mouthpiece using 3D-printing or CNC-milling technologies.
Almost all brass instruments during the MUSICES-project were scanned on a CT-facility which provides a 600 kV power supply. This high energy is necessary to irradiate material of such density. The
Fig.4
A cross section of the corner joints shows that not only thick parts like the lower block are not transmitted by the neutrons. In the X-ray image the bevelled joint of the corner is visible while the neutron image shows only artefacts (image errors) at the same location [Fig.2].
The painted decoration on the top plate is not shown in the neu-tron image. This is because of the metal pigments that were used, which cause minimal attenuation to the neutron beam and are therefore only visible in the X-ray image [Fig.3].
Despite all of its limitations in scanning wooden objects, the un-doubted advantage of neutron imaging is high contrast. For example, on
Fig.2
Fig.3
20 mm 20 mm
from lowest to highest intensity, from c. 15 pixels. Due to the setup of the CT-facility, the spatial resolution was limited to 118.4 μm voxel size. In this case it would not be easy to determine the exact surface, and therefore measurements, of the bore, for example, would always have this inaccuracy. An exact reproduction of the object that fulfils all acoustic requirements would be difficult. In the picture, it is also clearly visible that the distribution of the grey values are not homogenous. In thicker parts we find brighter areas than in thinner parts. The range of grey values in the diagram show a distribution of 100 units. Usu-ally, this would mean that there is material with a different attenuation coefficient. In this case the difference is due to hardening artefacts. The information about the metal structure is therefore unreliable.
The other mouthpiece of similar size was scanned with thermal neu-trons at the NEUTRA facility of the Paul Scherrer institute (Fig.5, top).
In the neutron CT-image, we see a much clearer outline of the object. In the diagram, the graph rises from zero intensity to highest intensity at a steeper pitch, and the range between the extremes on the edges is only 8 pixels. The strong differences in the diagram can be ex-plained in large part by noise. Determining the surface is possible with higher precision compared to the X-ray scan. Here a spatial resolution of 50 μm voxel size could be achieved. Therefore measurements can be conducted and a reproduction using 3D-printing or CNC-milling technologies would be more accurate.
2.5 Case Study 3: Scanning an Instrument with Mixed Materials
Scanning objects that consist of materials with highly differing densi-ties can be challenging. The different materials have varying attenua-tion coefficients. In the case of X-ray CT, problems occur for example when a wooden object has metal parts. This can be a nail in the neck of a violin or the keys of a wooden wind instrument. For the latter example, it could be a scientific issue to take measurements of the bore or to judge the conservational condition and visualise details like small cracks. Image errors such as artefacts can influence the image quality in a way that can make answering these questions difficult. For neutron imaging, the problem is the other way around. If wooden parts are too thick, they cannot penetrate.
In one experimental setup at the Paul Scherrer Institute, a flute with metal keys (German silver) was scanned twice. First, a CT scan was performed using thermal neutrons. Secondly, the object was scanned at the same position using X-rays without any prefiltering.
The images can now be compared [Fig.6].
Both images show a cross section of the flute where a big metal key is placed. The X-ray image shows only a bright shadow of the tube and a very bright area where the metal key is supposed to be depicted.
parameters. The X-ray spectrum is polychromatic, but can be filtered to use only a harder spectrum for a better irradiation. Here a 6 mm thick copper prefilter was used.
The cross section of the X-ray CT (Fig.5, bottom) shows on first sight a good outline of the object. The diagram visualises the distribu-tion of the grey values. On the bottom line we see the distance and on the left axis the intensity of the grey values at every point. It shows however, that the object outline is not very sharp. The left side of the graph (where the transition from air to object is visible) shows a range
Fig.5 Above: cross sections of two mouthpieces, below: distribution of grey values in the marked line (top: neutron-CT; bottom: X-ray CT).
Fig.5
The artefacts caused by the metal part are so strong that almost no de-tails from the wooden part or of the key’s construction are visible. The neutron image shows the constructional parts of the area in much more detail. The bright ring around the tube is caused by hardening artefacts, but details such as the angle of the tonehole opening or even the soft material between the tonehole and key are visible. Measuring the exact diameter of the bore is difficult because of the high noise. Depending on the organological or technological issue being explored, the neutron image provides much more information than the unfiltered X-ray CT.
As far as X-ray CT is concerned, a so called microfocus tube has to be used when high resolution is required. At the present time, this limits the possible tube voltage, with the result that metal artefacts will occur. In order to achieve a sufficient quality of mixed-material objects, a more elaborated scanning technique has to be used. With a combina-tion of different X-ray spectra, metal artefacts can be reduced.
During the MUSICES project, a couple of experiments for enhanc-ing the image quality for instruments containenhanc-ing of mix materials were undertaken [13]. The example shown here is a cor anglais (inv. no. MIR 396) which was scanned with two different spectra. First it was scanned with 220 kV tube voltage and a 0.89 mm Titan prefilter. In a second scan, a higher voltage of 225 kV was applied in combination with a 2.5 mm Cop-per prefilter. While the single reconstructions still show a lot of artefacts, the combination of both is capable to reduce them significantly [Fig.7]. 2.6 Case Study 4: Water Content in Musical Instruments
The following example and more quoted experiments shall outline the capacities of neutron imaging for the determination of water content in musical instruments. According to the experiments done by Ilona Stein [14] on the moisture content in woodwind instruments, the experimen-tal setup at the Paul Scherrer Institute should take advantage of the fact that a small water film can be made visible with neutron imaging.
Two sample tubes made of maple in a size of c. 73.3 mm length and 16.6 mm diameter were scanned dry, then wetted and scanned again.
One of the tubes was oiled beforehand with linseed oil; the other tube was not treated. To wet the tubes, a wet cloth was put into the bore for about ten minutes. The oiled tube absorbed only a small amount of water (0.06 g), and the untreated tube absorbed more (0.41 g). As the amount of water absorbed by the wooden (and thus highly attenuating) samples was relatively low, and due to the fact that most of the water evaporated from both tubes during the measurement, it was not possible
Fig.6
Fig.7
Fig.8
The fact that some materials are activated by the neutron beam can be a problem for the examination of cultural heritage objects and is thus another limiting factor. However, both techniques provide complemen-tary features for the various scientific issues on musical instruments.
ac k n o w l e d G e M e n t S: The research at the Paul Scherrer Institute was made possible by a STSM grant from the Cost Action FP1302 WoodMusICK programme (COST-STSM-FP1302-33359). I gratefully acknowledge the MU-SICES team, Ilona Stein in Nuremberg for providing two of her wooden test
ac k n o w l e d G e M e n t S: The research at the Paul Scherrer Institute was made possible by a STSM grant from the Cost Action FP1302 WoodMusICK programme (COST-STSM-FP1302-33359). I gratefully acknowledge the MU-SICES team, Ilona Stein in Nuremberg for providing two of her wooden test