The muon is a lepton with a mass of 105.7 MeV/c2, approximately 200 times heavier than the electron. The interaction between muons and materials has been used in various fields of research. Muon spin rotation, relaxation and resonance (μSR) use polarised positive muons as sensitive magnetic microprobes with short-range interactions with materials. Muon tomography uses the highly transmissive nature of muons to image low-density regions of materials. Recent studies on tomographic imaging with cosmic ray muons have succeeded in imaging the density structure of volcanoes1,2,3.

Muons entering materials lose their kinetic energy with less effective bremsstrahlung than electrons due to their heavier mass; they can also penetrate much deeper into materials than electrons. The penetration depth of muons depends on the density of the target material and the initial muon momentum. Negative muons are captured by atoms to form muonic atoms at the stopping depths. In a muonic atom, cascade transition of the trapped muon from higher to lower energy states occurs with the emission of characteristic muonic X-rays. As muons have an orbit closer to the atomic nucleus than electrons owing to its ~200 times heavier mass, the characteristic muonic X-ray has an energy ~200 times higher than that associated with electron transition (more than several ten keV even if the muon is captured in light elements). The muonic X-rays are thus intense enough to pass through the target material from the much deeper interior than the characteristic X-rays of electron transitions, thereby suggesting that muonic X-rays from muonic atoms can be applicable to non-destructive elemental analyses of the deep interior of materials, which cannot be performed with X-ray fluorescence spectroscopy or electron probe microanalysis.

In 1971, Rosen4 noted the great availability of muon beam analysis and proposed its application in the chemical analysis of tissues, as muon beam analysis would cause less damage to the host organism than neutron activation analyses. Muonic atom spectroscopy has been developed over the past four decades5,6,7 as a non-destructive analytical method; however, for a real application, one had to wait for an intense muon source. The intense pulsed muon source, J-PARC MUSE (Japan Proton Accelerator Research Complex, MUon Science Establishment), was constructed and succeeded in providing a decay muon rate of 106 s−1 for a momentum of 60 MeV/c in November 2009, which is the most intense pulsed muon beam in the world8,9. Although J-PARC was damaged by a magnitude 9 earthquake and a subsequent tsunami on 11 March 2011 (“Higashi-Nippon Dai-Shinsai10,11,12”), it was successfully revived in February 201213.

We report a feasibility study of the non-destructive muon beam chemical analysis of light elements (C, N, O and B) from millimetre- to centimetre-sized samples, which are difficult to analyse via neutron activation analysis, using the D2 muon beam line at J-PARC MUSE. Moreover, the degree of radioactivation due to muon irradiation is much less than that of neutron activation analysis, thereby allowing various applications. A potential useful application of this technique is, for instance, non-destructive analyses of light elements in extraterrestrial samples, particularly those obtained in sample return missions. Two sample return missions, Hayabusa-214 and OSIRIS-REx15, will aim for C-type asteroids, for which reflectance spectra resemble those of carbonaceous chondrites containing organic materials. Millimetre- to centimetre-sized asteroidal surface samples will be returned in the 2020s14,15 and the non-destructive analyses of light elements will prove to be a powerful analytical tool to determine the contents and/or distribution of organic materials, which may have been prebiotic building blocks of life, in pebble-sized noble samples.


The depth profiling of light elements from a layered sample

The muon beam analysis of a four-layered sample consisting of SiO2 glass, graphite (C), boron nitride (BN) and SiO2 glass was generated to obtain a depth profile of light elements. The sample is held by an aluminium holder in an aluminium vacuum chamber with a beam-entrance window composed of polyimide foil.

The negative muon beam was collimated to approximately 2.7 cm in diameter and focused on the sample surface (50 mm × 75 mm), which was oriented at 45 degrees to the beam (Fig. 1). Adjusting the muon momentum from 32.5 MeV/c to 57.5 MeV/c, muonic X-rays from the sample were measured by a Ge detector. The accumulation time of X-ray signals was 3–4 hours for each muon momentum. Figure 2 shows the observed X-ray spectra for the different momentum of incident muon beams. The energies of muonic X-rays listed in Engfer et al.16 were used for peak identification. Note that the peaks at 66 and 89 keV represent Al signals from the sample holder and vacuum chamber. At the momentum of 32.5 MeV/c, the µSi-Lα (76 keV) and µO-Kα (133 keV) signals were detected with the background µAl-Lα (66 keV). The peaks of µSi-Lα and µO-Kα were also confirmed at the momentum of 37.5 and 40.0 MeV/c, although they disappeared at the momentum 42.5 MeV/c, where the µC-Kα (75 keV) signal was clearly observed. As the µC-Kα peak should come from the second graphite layer, these observations suggest that the muon beam penetrated through the first SiO2 glass layer and into the depth of >2 mm of the sample at the momentum of 42.5 MeV/c. The findings also demonstrate that the characteristic muonic X-ray of carbon was emitted from a depth of >2 mm through the overlying SiO2 glass layer without significant self-absorption and emission of muonic X-rays of Si and O. At the momentum of 50.0 MeV/c, the µC-Kα peak disappeared and only µB-Kα (52 keV) and µN-Kα (102 keV) signals from the third layer (boron nitride) were detected through the overlying SiO2 and graphite layers. Finally, at the momentum of 57.5 MeV/c, the peaks of µB-Kα and µN-Kα disappeared and the µSi-Lα (76 keV) and µO-Kα (133 keV) signals were again detected. This clearly shows that the muon beam penetrated into the depth of >5.3 mm through the layered sample and the µSi-Lα (76 keV) and µO-Kα (133 keV) from muonic atoms in the fourth SiO2 glass layer were emitted through the overlying layers.

Figure 1
figure 1

Geometry of the depth profile analysis.

A four-layered sample (SiO2-graphite-BN-SiO2) and a Ge detector were oriented at 45 and 90 degrees to the negative muon beam, respectively, in an aluminium vacuum chamber with a polyimide window. The muon beam was collimated to 4-2.5 cm in diameter. The sample was supported by a holder made of aluminium foil.

Figure 2
figure 2

Muonic X-ray energy spectra from the four-layered sample.

Muonic X-ray intensities (y-axis) are shown in arbitrary units with offset to compare the spectra obtained with different muon momenta (32.5–57.5 MeV/c). The peaks at 66 and 89 keV are attributed to Al used for a sample holder and the vacuum chamber. The peaks of µSi-Kα and µO-Kα were detected from the first SiO2 glass layer at the muon momentum of 32.5–40.0 MeV/c. The muonic X-ray of C (µC-Kα) from the graphite layer and those of B and N from the BN layer were detected at the momenta of 42.5–47.5 and 47.5–52.5 MeV/c, respectively. The peaks of µSi-Kα and µO-Kα were detected again at the muon momentum of 55.0–57.5 MeV/c from the bottom SiO2 glass layer through the three overlying layers.

The intensities of muonic X-ray signals of B, C, N, O and Si with 1σ uncertainties from counting statistics of X-rays are plotted against the muon momentum, as displayed in Figure 3. Because the contributions of µAl-Kα from the vacuum chamber and the sample holder should be almost constant irrespective of the muon momentum, the intensity of each peak of muonic X-ray was normalised to that of µAl-Lα. The stopping depth of muons in the sample at a given momentum was estimated based on the Bethe-Bloch formula, which is generally used to calculate the energy loss of charged particles travelling in a medium, using the density and thickness of each layer. The depths corresponding to the boundaries of sample layers (indicated by arrows) are consistent with the change in muonic X-ray signals from the sample.

Figure 3
figure 3

The depth-profile of the four-layered sample.

The muonic X-ray intensities of B, C, N, O and Si from the layered sample are plotted against the muon momentum. The muonic X-ray intensity of each element is normalised to the intensity of μAl-Kα at each momentum and to its maximum count along the depth profile. Analytical uncertainties (1σ) are estimated from the counting statistics of X-ray counts because the contributions of uncertainties in X-ray detection efficiency are much smaller than those of counting statistics (typically less than 5%). The stopping depth of muons at a given momentum is estimated from the Bethe-Bloch formula and the expected layer boundaries are indicated by arrows.

The self–absorption effects on µSi-Lα (76 keV) and µO-Kα (133 keV) at the depth of ~7 mm are estimated to be 20% of the densities and thicknesses of overlying layers, which are consistent with slightly weak peak intensities of µSi-Lα (76 keV) and µO-Kα (133 keV) at the momenta of 55 and 57.5 MeV/c compared with those of the first layers (32.5 and 37.5 MeV/c) (Fig. 3).

The current depth estimate should have a ~10% uncertainty largely due to the momentum range of the incident muon beam. Although there are further developments in experimental techniques, such as momentum filtering of the incident muon beam and a position-sensitive detector, we currently conclude that the non-destructive depth profile of light elements was successfully obtained up to ~7 mm in depth with a depth resolution of sub-mm. This result is hardly possible with other analytical techniques such as neutron activation analysis, X-ray fluorescence spectroscopy and electron probe microanalysis. Furthermore, muon beam analysis displays significant potential for the analysis of light element distributions within millimetre- to centimetre-sized samples.

Elemental analysis of meteorite samples

Muonic beam analysis was applied to primitive meteorites called chondrites. Chondrites have never been melted and remain physical aggregates of various components that formed in the early solar system prior to planet formation. Carbonaceous chondrites are one of the various chemical groups of chondrites, some of which contain abundant water as hydrated minerals and organic materials17. They record the long evolutionary history of the solar system and prebiotic organic reservoirs. Thus, future asteroidal sample return missions (Hayabusa-2 and OSIRIS-REx) plan to sample such materials from asteroids without terrestrial contamination and with geologic contexts14,15.

In this study, we carried out a non-destructive muon analysis of chips of carbonaceous chondrites, Murchison and Allende. The Murchison meteorite contains a copious amount of diverse extraterrestrial organic materials, while the Allende meteorite contains less carbon than the Murchison meteorite.

The negative muon beam, collimated to approximately 4 cm in diameter, irradiated to a ~5-mm-thick Murchison disk with the exposed surface area of ~5 cm × ~10 cm. The meteorite disk and the Ge detector were oriented at 45 and 90 degrees to the negative muon beam, respectively, in the same manner shown in Fig. 1. Aluminium foil was used to hang the sample inside the chamber. The exposure time was approximately 13 hours and the momentum of the incident muon beam was set at 16 MeV/c, corresponding to the penetration depth of ~70 μm. For Allende meteorite analysis, the muon beam was collimated to approximately 2.7 cm in diameter at the sample surface of 50 mm × 75 mm. The sample was also hung with less amounts of Al foil than for the Murchison meteorite to reduce the background signal of Al. The exposure time was approximately 10 hours and the incident muon momentum was set at 34 MeV/c (the penetration depth of ~1 mm).

Fluorescent X-rays of Mg, C, Si, Fe, Ca and S were detected from the Murchison meteorite, while X-rays of Mg, Si, Fe, K and Ca were detected from the Allende meteorite (Fig. 4; Table 1). Clear signals of Al were also obtained, although they should include background signals from the vacuum chamber and/or the sample holder and are therefore excluded from further discussion. Taking the emission probabilities of Kα and Lα fluorescent lines and the energy-dependence of Ge detector efficiency into account16, the captured muon counts for C, Si, Mg and O and for S and Fe were obtained from the integrated intensities of the Kα and Lα fluorescent lines, respectively. In Fig. 5, the captured muon counts normalised to that of Mg are plotted against the Mg-normalised weight concentrations of elementsin the Murchison and Allende meteorites. The captured muon count is roughly proportional to the weight concentration, with the exception of O, which has a high muon capture probability due to its high electronegativity18. The linear correlation in Fig. 5 indicates that the muon capture rate of an element is largely controlled by its concentration in a meteorite sample and neither self-absorption of muonic X-rays with different energies nor difference in chemical bonding have significant effects on muon capture.

Table 1 Characteristic muonic X-ray counts from carbonaceous chondrites, the Murchison and Allende meteorites
Figure 4
figure 4

Muonic X-ray energy spectra from carbonaceous chondrites: Murchison and Allende meteorites.

Fluorescent X-rays of Mg, C, Si, Fe, Ca and S were detected from the Murchison meteorite (blue), while those of Mg, Si, Fe, K and Ca were detected from the Allende meteorite (red). The peaks of Al includes signals from the vacuum chamber and/or the sample holder. The peak of µC-Kα (75 keV) was detected from the Murchison meteorite, which contains extraterrestrial organic matter with the total carbon of a few weight percent. The muon beam momenta of 16 MeV/c (penetration depth of ~70 µm) and 34 MeV/c (penetration depth of ~1 mm) were used for analysis of the Murchison (13-hour exposure) and Allende (10-hour exposure) meteorites.

Figure 5
figure 5

Correlation between the elemental concentration and the captured muon counts.

The captured muon counts by each element, normalised to those of Mg, are plotted against the Mg-normalised weight concentrations of elements in meteorite samples33 after correction of the energy-dependent detector efficiency and the emission probabilities of the Kα and Lα fluorescent lines. The captured muon counts correspond to the characteristic muonic X-ray counts. Error bars show analytical uncertainties (1σ) estimated from the counting statistics of X-ray counts.

Elements with concentrations of more than 1 weight percent were detected in this study. The detection limit of ~1 weight percent is attributed to the inefficient detection of muonic X-rays with a single Ge detector. Only 1% of the total emitted muonic X-rays can be detected by the Ge detector (5 cm in diameter) with the current experimental set-up, which will be significantly improved, i.e., with multiple detectors covering a much larger solid angle.

We also attempted to measure much smaller amounts of meteorite samples inside glass tubes to simulate non-destructive analyses of future return samples. Sealing extraterrestrial samples inside glass tubes was originally planned for samples from the asteroid Itokawa. Although Itokawa particles were not sealed in glass tubes due to their small sizes19,20,21, sealing in a glass tube is one of the effective ways to avoid terrestrial contamination of organic materials and volatiles and thus could be used in future sample return missions.

Powdered Murchison meteorite (610 mg) was placed in a 5-cm-long SiO2 glass tube, in which the inner and outer diameters were 4 mm and 6 mm, respectively. The muon beam collimated to approximately 2.5 cm in diameter and the apparent cross section of the sample was 4 mm × 25 mm. After exposure of the muon beam with the momentum of 37 MeV/c for approximately 24 hours, clear signals of Mg and marginally resolved signals of Fe were detected through the 1-mm thick glass wall (Fig. 6). Although O and Si are the major elements of rock samples, muonic X-rays of O and Si were emitted from the SiO2 glass tube as well, which cannot be distinguished from the sample signals. Although further developments in analytical techniques are required, such as detector setting and collimation of the incident muon beam, our first attempt to non-destructively measure an extraterrestrial sample inside a glass tube succeeded with the detection of Mg and Fe.

Figure 6
figure 6

Muonic X-ray spectra from the powdered Murchison meteorite in an SiO2 glass tube.

A clear signal of Mg and a marginally resolved signal of Fe from the sample were detected through the 1-mm thick SiO2 glass wall. The intense signal of Si originated from the SiO2 glass tube. The muon beam momentum was set at 37 MeV/c and the exposure time was 24 hours.


This study demonstrates that muon beam analysis is feasible for the non-destructive elemental analysis of light elements in extraterrestrial samples. This technique is complementary to other non-destructive analytical techniques, such as synchrotron X-ray microtomography and synchrotron X-ray diffraction, both of which have been used for preliminary examination of Itokawa particles19,20 to obtain structural and mineralogical information. Neutron activation analysis was also applied to Itokawa particles for bulk elemental analyses, including trace elements21. X-ray photoelectron spectroscopy (XPS)22,23 and X-ray absorption spectroscopy (XAS)24 at synchrotron facilities have been used to identify the chemical bonding of elements. Synchrotron X-ray absorption near-edge structure (XANES) spectroscopy has been used to identify chemical bonding and functional groups in cometary organic matter sampled from the comet 81P/Wild 2 by the Stardust spacecraft25. Nuclear reaction analysis (NRA) can be used to determine the contents and distribution of C, N and O in minerals26. The recent development of direct tomography with chemical bond contrast using synchrotron X-ray has made it possible to make a 3D map of the chemical bonding of light elements in a cm-sized sample with a spatial resolution of ~100 μm27. These techniques have not only benefits but also disadvantages, including severe radioactivation of samples, limitations in sample size and analytical area and/or depth, limited number of measurable elements and difficulties in bulk analysis. Muon beam analysis is capable of performing a bulk analysis of light to heavy elements at one time without severe radioactivation and is thus a unique and complementary analytical method.

J-PARC MUSE plans to increase the power of the proton beam, of which irradiation of a graphite target produces the muon beam, up to 1 MW13. The muon beam intensity will increase to ~107 count/s (i.e., ten times larger than the current beam), which will yield a significant improvement in the counting statistics of muonic X-ray detection. The counting statistics will also be improved by development of the detection system (e.g., multiple Ge detectors covering a larger solid angle28). With a more effective detector setting, the chemical structure states of Fe may be determined because the muonic X-ray structure, such as μKα/μKβ ratios, could depend on chemical structure state29,30. Iron is present as metallic Fe-Ni, iron sulfides, FeO in silicates and chondrites and its oxidation state has long been used as one of the classification criteria of chondrites. Moreover, this method has a future prospect of non-destructive three-dimensional mapping of multi-elements using, for instance, a position-sensitive detector with a large cross-section, such as Compton Camera31,32 and may represent the dawn of another “eye” after the proposition of muon radiography by Rosen4 more than forty years ago.


The muon science facility (MUSE) is located in the Materials and Life Science Facility (MLF), a facility that includes both neutron and muon science programs, at the Japan Proton Accelerator Research Complex (J-PARC) in Tokai, Japan. MUSE currently has two beam lines to extract surface muons and decay muons from the muon source target into the experimental halls13. Negative decay muons (5–120 MeV/c), which are obtained via the in-flight decay of µ+ confined by a strong longitudinal magnetic field of a superconducting solenoid magnet, were used in this study. Since November 26, 2010, the proton beam power from the Rapid Cycling Synchrotron has been steadily increased up to 220 kW, consequently delivering approximately 106 count/s of decay muon beam (µ+ and µ), until the earthquake on March 11, 2011. Although MUSE was seriously damaged from so-called ‘Higashi-Nippon Dai-Shinsai’, it recovered in six months and restarted operations for external users in February 2012. As a top priority analysis, this study was carried out in February 2012 and January 2013. The muon beam line at J-PARC/MUSE is now open for external users through a peer-reviewed proposal system (twice a year) as other analytical techniques in synchrotron facilities.