Introduction

Modification of materials using swift heavy ions (SHIs), i.e., high energy ions with stopping power in materials that are primarily dominated by electronic forces, has recently received an increasing amount of attention1. The examples include the anisotropic deformation of amorphous metals/glasses2,3, shape elongation of metal nanoparticles (NPs) embedded in materials4,5,6,7,8,9,10,11,12,13,14,15, and etched ion-track engineering16. When NPs embedded in SiO2 are irradiated with SHIs, the NPs are elongated toward the direction parallel to the SHI beam. Many impacts of SHIs gradually transform originally-spherical NPs to nano-rods. This paper describes that the same deformation is also induced by 1–6 MeV fullerene (C60) ions, while such C60 ions are not classified in SHIs because of their slow velocities. The energy per nucleon of several MeV C60 ion, which is an index of ion velocity, is on the order of 10−3 MeV/u, while that of SHI is ~1 MeV/u or higher. However, C60 ions provide high electronic energy deposition comparable to SHIs due to the coincident impacts of 60 carbon atoms within a limited area of the C60 molecule size of ~0.7 nm in diameter. The stopping powers of the C60 ion is approximated as the sum of sixty independent carbon monomer ions with the energy of each one (E/60) expressed as,

$${S}_{i}(E,\,{{\rm{C}}}_{60})=60\,{S}_{i}(E/60,\,{{\rm{C}}}_{1})$$
(1)

where i = n (nuclear) or e (electronic)17. The monomer stopping power Si (E/60, C1) is derived from SRIM 2013 code18. A 6 MeV C60 ion provides ~15.5 keV/nm of electronic stopping power Se in amorphous SiO2, which is as high as Se of 200 MeV Xe ion (15.0 keV/nm), as shown in Table 1. In this paper, C60 ions of 1–6 MeV were applied, which correspond to Se of 6.3–15.5 keV/nm. Using the same Se range of SHIs, we have already succeeded in the elongation of NPs as shown in Table 14.

Table 1 Electronic and nuclear stopping powers (Se, Sn) at the surface of amorphous SiO2 calculated from SRIM2013 code for SHIs and from the Eq. (1) for cluster ions. The elongation efficiency is defined as the inverse of the fluence where the optical dichroism becomes 0.02.
Full size table

Since 1990s, C60 ions of up to 40 MeV have been available at the Orsay facility, France19,20,21. However, the available fluences have been limited because of the extremely low fluxes on the order of 106 C60/cm2s21. Conversely, obvious NP elongation was thought to require high fluences as 1013 C60+/cm2 or more, i.e., irradiation time of ~4 months. The elongation experiments have only recently become possible because the Takasaki group, a part of the research team that conducted this study, has developed an electron-attachment-type C60 ion source which generates enormous fluxes of ~1010 C60/cm2s and consequently attains the high fluences22.

Using the high-flux C60 ion source, we have firstly succeeded in elongating the shape of NPs by the cluster ion irradiation. Simultaneously we observed different phenomena which have not been observed under SHI irradiation, i.e., the formation of short ion tracks, and significantly enhanced sputtering, and unexpectedly weak velocity effect. In this paper, the interplay between these phenomena and NP elongation are also discussed.

Results

Elongation of NPs: TEM observation

Figures 1(a–d) show bright field (BF) cross-sectional TEM (XTEM) images of Au NPs embedded in SiO2, (a)(c) before and (b)(d) after the irradiation with 4 MeV C60+ ions to a fluence of 5 × 1013 C60+/cm2. The incident angle of the C60+ beam was set to 45°, as shown by the arrows in Figs. 1(b,d). Before the sample thinning, a thin Pt layer was deposited as a surface marker. In unirradiated state (Figs. 1(a,c)), a monolayer of Au NPs was observed, each of which exhibited an oblate spheroidal shape with principle axis pointing to the surface normal (see supplementary materials). The Pt marker was observed on the top of a 100 nm thick SiO2 layer over the Au NP layer as shown in Fig. 1(c). After the irradiation with 5 × 1013 C60+/cm2 (Fig. 1(d)), the thickness of the deposited SiO2 layer markedly decreased from ~100 nm to ~30 nm due to enhanced sputtering associated with the C60 irradiation. Consequently, the distance between the Pt marker and Au NPs considerably decreased, which could make it difficult to distinguish between the markers and NPs, though scanning TEM and X-ray energy-dispersive spectrometry (STEM-EDS) mapping made it possible to clearly distinguish the Pt marker and Au NPs, as shown in Figs. 1(e,f).

Figure 1
figure 1

Cross-sectional TEM images of Au NPs embedded in SiO2 before (a,c,e) and after (b,d,f) irradiation with 4 MeV C60 ions at a fluence of 5 × 1013 C60+/cm2. Enlarged images of Au NPs (a,b) and overviews of deposited SiO2 layer on the NPs (c,d). A thin layer of Pt was deposited for surface marker. (a–d) Bright field (BF) images and (e,f) STEM-EDS element mappings. In (e,f), red, blue, and green regions correspond to Pt, Si, and Au rich regions.

Full size image

The widely distributed markers in Fig. 1(d) manifested a very rough surface caused by the sputtering. Figures 1(b,d) clearly shows that the Au NPs raised the major axes toward a 45° angle, which was the incidence angle of the ion beam. In addition, the NPs changed the shapes to nearly prolate spheroids with the principle axes pointing in the direction of the 45° beam angle. These observations have clearly indicated that the shape elongation of NPs was induced by the slow and low energy 4 MeV C60+ ions.

Elongation of NPs: optical dichroism

Optical absorption spectra of Zn NP samples23,24 irradiated with C60 ions to various fluences were detected under linearly polarized light illumination. The polarization angle is 0° when the polarization plane includes the major axes of the elongated NPs. As reported in ref.4,11, the difference in the optical absorption at 0° polarization and 90° polarization is, in certain conditions, proportional to the aspect ratio (AR) of the elongated NPs. The absorption spectra at 0° (solid curves) and 90° (broken curves) polarization at various fluences are presented in Fig. 2(a,b) for 6 MeV (Se = 15.5 keV/nm) and 2 MeV (9.3 keV/nm) C60 ions, respectively. At both the ion energies, the Se was high enough to induce NP elongation if it was supplied from SHIs as shown in Table 1.

Figure 2
figure 2

(a,b) Optical absorption spectra of embedded Zn NP samples detected by linearly polarized light, which were irradiated with 6 MeV and 2 MeV C60+ ions at various fluences indicated in the figures. The transmittance of light is denoted by T. The light polarization angle was 0° (solid curves) when the polarization plane included the major axes of the NPs. (c) Fluence dependence of Zn content in the samples irradiated with C60+ ions of 2 MeV (triangles), 4 MeV (open circles), and 6 MeV (closed circles), detected by RBS. (d) Energy dependence of nuclear stopping power (Sn) and electronic stopping power (Se) of C60 ions in SiO2, calculated by SRIM 2013 code using Eq. (1). (e,f) Photoimages of Zn NP samples irradiated with 6 MeV C60+ ions at a fluence of 1 × 1011 C60+/cm2 and 5 × 1013 C60+/cm2, respectively. The dark brown color is due to Zn NPs dispersed in SiO2. A colorless region in the sample (f) indicated that Zn NPs were mostly lost due to sputtering by C60 irradiation, which was also confirmed by RBS.

Full size image

The dependence between 2 and 6 MeV ions is qualitatively similar but not quantitatively. In an unirradiated state, both spectra detected at 0° and 90° polarization fall on the same curve. However, as the fluence increased, spectral deviation between 0° and 90° polarization, i.e., the difference between the solid and broken curves increased up to 1 × 1013 C60+/cm2. Simultaneously, the absorption peak intensity at ~4.8 eV decreased with increased fluence, which can be attributed to sputtering loss of Zn atoms from the samples. The absorption mostly disappeared at 5 × 1013 C60+/cm2 for 6 MeV but a half of the peak remained for 2 MeV. According to Rutherford backscattering spectrometry (RBS) measurements combined with Rump code analysis25, more than 90% and 40% of Zn atoms were lost from the samples at the fluence of 5 × 1013 C60/cm2 with 6 MeV and 2 MeV C60 ions, respectively (Fig. 2(c)). Figures 2(e,f) show images of the samples irradiated with 6 MeV ions at 1 × 1011 and 5 × 1013 C60+/cm2. While Zn NPs in SiO2 show brown color (Fig. 2(e)), a colorless region was clearly observed at 5 × 1013 C60+/cm2, as shown in Fig. 2(f), indicating the severe loss of Zn NPs caused by sputtering. Samples irradiated with 2 MeV C60 ions displayed lower efficiencies with respect to both the NP elongation and the sputtering loss.

The fluence dependence of NP elongation was determined from optical spectra using the method described in4 and is presented in Fig. 3(a), with the SHI data of 200 MeV Xe ions for reference. In the low fluence region between 1 × 1011 and 1 × 1012 C60+/cm2, 4 MeV (Se = 12.7 keV/nm) and 6 MeV (15.5 keV/nm) C60 irradiation induced shape elongation comparable to or slightly higher than 200 MeV Xe irradiation (15.0 keV/nm), while irradiation at 1 and 2 MeV induced less elongation. However, it should be noted that the elongation is induced with even 1 MeV C60 ions (Se = 6.3 keV/nm). It is not surprising because the 1st threshold of the shape elongation by SHIs is ~3 keV/nm4. The 1 MeV C60 ions have higher Se than the threshold. It is noted that the elongation is induced by SHIs with much lower efficiency even below the 1st threshold4. The slightly higher efficiency of 4 and 6 MeV C60 ions compared to 200 MeV Xe ions could be attributed to two: (i) The much slower velocity of the cluster ions compared to SHIs, which reduces the energy of δ-rays and excitation volume, resulting in a higher excitation density26. (ii) The synergy effect between Sn and Se. While Sn is much lower than Se for SHIs, this is not the case for MeV C60 ions, as shown in Table 1. Consequently the synergy effect is not excluded.

Figure 3
figure 3

Elongation of Zn NPs embedded in SiO2 was determined by optical dichroism spectroscopy. (a) Fluence dependences of the elongation induced by Xe ions of 200 MeV, and C60 ions of 6 MeV, 4 MeV, 2 MeV, and 1 MeV, respectively. A broken line indicates the slope of unity. (b) Se dependence of the elongation efficiencies of Zn NPs for C60 ions of 1, 2, 4, and 6 MeV, i.e., E/M < 0.009 MeV/u (open circles) and for swift heavy ions (>1 MeV/u), i.e., 200 MeV Au13+, 200 MeV Xe14+, 60 MeV Ti6+, and 50 MeV Si5+ (closed circles). (c) The elongation is plotted against the product of the electronic stopping power and the fluences, i.e., the electronic deposited energy density.

Full size image

From the slope of unity in the log-log plot in Fig. 3(a), the NP elongation linearly increased with the fluence up to 1 × 1012 C60+/cm2. However, the slope was reduced to less than unity above 1 × 1012 C60+/cm2, and finally turned to a negative slope around 1 × 1013 C60+/cm2. The sublinear increase in intermediate fluences and the decrease in high fluences can be attributed to the destruction of the elongated NPs caused by the enhanced sputtering of the cluster ions. In fact, the quantities of Zn atoms in the samples decreased with the fluence as evidenced by RBS, as shown in Fig. 2(c).

To compare the elongation efficiency between the different beams, the efficiency was defined as the inverse of the fluence, where the optical elongation reaches to the value of 0.02, and shown in Table 1 and Fig. 3(b). See details for ref.4. (While the fluence at which the optical elongation reached to 0.10 was used in ref.4, the fluence at which the elongation reached to 0.02 was used in this article, because the linear region was limited.)

As shown in Fig. 3(b), the maximum efficiency was obtained under 4 MeV C60 ion irradiation. However, both the data by SHIs (closed circles) and by C60 ions (open circles) fall on almost the same dependence, indicating that large difference due to the velocity effect was not observed.

In the previous literature, the elongation induced by SHI irradiations is normalized when the abscissa is plotted with the product of Se and the fluence Φ. Therefore the elongation is determined only by the electroni deposited energy density4. Of course, this kind of simple relation is not applied to sputtering induced by SHIs27. To check whether this simple relation is hold in this case or not, the optically detected elongation was plotted with the product of Se and the fluence Φ. In contrast to SHIs, the data points do not fall on the same curve. A simple relationship like the elongation induced by SHI irradiation was not observed, probably this process included the strong electronic sputtering.

Ion track formation

Although the mechanism of NP shape elongation induced by SHIs is yet to be identified, a majority of researchers agree with the importance of the ion-track formation in matrix materials4,5,6,7,8,9,10,11,12,13,14,15. However, the tracks cannot be directly observed by TEM in amorphous matrices28, except in a very thin self-standing film29. For reference, a piece of crystalline SiO2 (c-SiO2) without NPs was irradiated with the C60 ions and observed by XTEM. Figure 4(a) presents an XTEM image of the sample irradiated with 4 MeV C60 ions at a fluence of 5 × 1011 C60+/cm2. The incident angle of the C60 beam was set to 7° from the surface normal to avoid channeling. The irradiated surface was coated by a thin Pt layer as a surface marker for XTEM analysis. It should be noted that the surface was much flatter than that in Fig. 1(b) because the fluence was much lower.

Figure 4
figure 4

(a) A bright field cross-sectional TEM image of c-SiO2 irradiated with 4 MeV C60+ ions at a fluence of 5 × 1011 C60+/cm2 with an incident angle of 7° from the surface normal. A thin layer of Pt was deposited as a surface marker after irradiation. The inset shows an expanded image. (b) Depth dependence of C60 ion energy injected into crystalline SiO2 with incident energy of 4 MeV (lower) and 6 MeV (upper), calculated from Eq. (2). Depth dependences of electronic stopping power Se and of total stopping power, Se + Sn, are plotted by broken lines and chain lines, respectively. Experimentally observed track lengths, and the threshold Se for track formation in c-SiO2 of Seth = 2 keV/nm30, are also shown in the figures. (c) The effects of a deposited SiO2 layer of 200 nm thick on a Zn NP sample are monitored by linearly polarized absorption spectra. The spectra (i) were detected from the Zn NP sample with the deposited layer, which were irradiated with 6 MeV C60 ions to 1 × 1013 C60+/cm2. In this configuration, the ion tracks do not reach the Zn NP layer, while calculations estimate that the Zn NPs could be irradiated with ~2.5 MeV C60 ions through the deposited layer if the C60 ions would not be fragmented. The spectra (ii) and (iii) were detected from the Zn NP samples without the deposited layer, irradiated with 2 MeV and 6 MeV C60 ions, respectively. Polarization angle of 0° (solid curves) is defined when the polarization plane includes major axes of NPs, while that of 90° (broken curves) is defined when the plane is perpendicular to the axes.

Full size image

Many cylindrical structures of 10.4 ± 3.0 nm in diameter and 60–80 nm in length were observed below the black surface marker layer in Fig. 4(a). The observed track length of 60–80 nm was unexpectedly short. It was experimentally observed that the track length increased when the ions were irradiated exactly at the surface normal, i.e., 0°, probably due to channeling. The energy of the C60 ion at depth x in SiO2 was calculated based on the rigid C60 molecule model, i.e., the fragmentation of C60 molecule is excluded,

$$E(x,{{\rm{C}}}_{60})={E}_{o}-{\int }_{0}^{x}[{S}_{e}(E,{{\rm{C}}}_{60})+{S}_{n}(E,{{\rm{C}}}_{60})]dx,$$
(2)

where Eo is the incident energy of the C60 ion. The results for 4 and 6 MeV ions are shown in Fig. 4(b). The stopping powers Se(E) + Sn(E) and Se(E) are provided in Fig. 4(b). Since the threshold electronic stopping power for track formation in c-SiO2 is known to be ~2 keV/nm30, tracks by 4 MeV C60 ions would be formed down to a depth of 270 nm, where Se > 2 keV/nm. It should be noted again that the experimentally observed track lengths were 60–80 nm, much shorter than the rigid model predicted.

As more clearly shown in the inset of Fig. 4(a), dot-like structures are observed on the vacuum side of the surface marker layer. Judging from the intensity of the dot images, they are likely from Pt, but were probably pushed toward the vacuum side by the hillocks of c-SiO2 underneath. The hillocks are indirectly observed in this case. In fact, it is clearly observed that the lateral dimensions of the dots are larger the diameters of the tracks inside of c-SiO2, because of the thickness of the deposited Pt marker. The hillocks are most likely formed when the tracks met the surface31.

Shape elongation deeper than the track lengths

The next question is whether NP elongation was only induced within the track depth or also beyond. A 200 nm thick layer of amorphous SiO2 was deposited on a Zn NP sample, irradiated with 6 MeV C60 ions, and compared with samples without the deposited layer. As shown in Fig. 4(c), a weak hump was observed at ~2.6 eV for a sample with the deposited layer (curves (i)), which was attributed to an interference fringe in the deposited layer.

The 200 nm thickness was wider than the typical track length of ~120 nm for 6 MeV C60 ions. In this configuration, Zn NPs did not interact with the ion tracks unless the deposited layer was severely sputtered. According to Fig. 4(b), the mean energy of C60 ions was reduced from 6 MeV to ~2.5 MeV after passing through the 200 nm thick SiO2 layer. The energy of ~2.5 MeV was still high enough to induce NP elongation since the anisotropic absorption, i.e., elongation was observed even under 2 MeV C60 irradiation, as shown by curve (ii) in Fig. 4(c) and also Fig. 2(b). It should be noted that the direct irradiation of the Zn layer with 2 MeV C60 ions induced higher elongation than indirect irradiation with ~2.5 MeV C60 ions through the 200 nm-thick deposited layer, although the latter had higher energy. This observation can be interpreted that the 6 MeV C60 ion lost the energy going through the deposited layer of 200 nm thick and was fragmented. The fragmented carbon ions which totally have ~2.5 MeV no longer to form ion tracks nor elongation of NPs. Probably the distances between the fragments become too long to induce the cooperative effect between them. However, since the track length of ~120 nm was determined from quartz but not from amorphous SiO2, further confirmation is necessary.

Discussion

Enhanced sputtering

It is well known that there are two types of sputtering mechanisms: nuclear and electronic sputtering. Furthermore, the synergy effect between both types could be possible32. Since the present case is in an intermediate energy region, we evaluated which mechanism, i.e., Sn or Se was dominant for the observed enhanced sputtering under C60 irradiation.

Energy dependences of stopping powers, Se and Sn, of C60 ions are plotted in Fig. 2(d), indicating that Se was higher than Sn when E > 1 MeV. However, Sn was not negligible, which is different from SHIs. An important observation was revealed in this energy region, namely Se increased but Sn decreased with ion energy. Figure 2(c) shows the fluence dependence of the Zn loss induced by C60 ion irradiation at three different energies, 2, 4, and 6 MeV. The Zn loss was more efficient at higher energies. The energy dependence of the sputtering efficiency and that of Se and Sn are presented in Table 2. The electronic stopping power Se and the observed sputtering yield increased with increasing the energy, while the nuclear stopping power Sn decreased. This difference in energy dependence indicated that sputtering can be attributed to an Se-related mechanism, not an Sn-related one.

Table 2 Energy dependence of sputtering efficiency (observation) and of electronic and nuclear stopping powers calculated.
Full size table

A further quantitative evaluation is presented in Table 3. The sputtering yield of SiO2 from C60 ion irradiation was estimated using both the models. In the nuclear sputtering model, the nuclear sputtering yield Yn by a C60 ion with energy E is given by a sum of those independent sixty C monomer ions with E/60,

$${Y}_{n}(E,{{\rm{C}}}_{60})=60{Y}_{n}(E/60,\,{{\rm{C}}}_{1})$$
(3)

The sputtering yield of SiO2 by the C monomer ion, Yn (E/60, C1), was determined by SRIM 2013 code. The yield was 16.3 atoms/C60-ion for 6 MeV C60 ions.

Table 3 Comparison of the experimental sputtering yield with the nuclear and the electronic sputtering models.
Full size table

In the electronic sputtering model, a C60 ion, which is characterized as Se, is treated as a swift heavy ion. Applying an experimental cubic formula for SiO2,

$${Y}_{e}={B}_{1}{S}_{e}^{3},$$
(4)

where B1 = 0.20 atoms/ion (nm/keV)3 27, the yield of 745 atoms/C60-ion was estimated for 6 MeV C60 ion. As shown in Fig. 1, the thickness of the SiO2 top-layer decreased from 100 to 30 nm under 4 MeV C60 irradiation at a fluence of 5 × 1013 C60+/cm2. Also as shown in Fig. 2(f), the 70 nm of thick Zn NP layer was completely sputtered out under 6 MeV C60 irradiation, below 5 × 1013 C60+/cm2. The observed sputtering yields were 3.2 × 103 and >3.2 × 103 atoms/C60 for 4 MeV and 6 MeV irradiation, respectively, both of which were much higher than the calculated values. The sputtering yield was greatly enhanced probably due to cluster enhancement or the synergy effect17,33,34.

Track diameter

As shown in Fig. 4(a), the track diameter of 10.4 ± 3.0 nm was determined in crystalline SiO2 under 4 MeV C60+ irradiation (Se = 12.7 keV/nm). Afra et al. summarized various experimental data of Se dependence of the track sizes of crystalline SiO2 with the i-TS calculations of the velocities of 0.5 MeV/u and 7.0 MeV/u30. The calculated track diameters for 15.3 keV/nm (4 MeV C60+ in c-SiO2) were ~11 nm and ~9 nm for the low and high velocities, respectively. The track radius of the 4 MeV C60+ ion in c-SiO2 was not so large compared to SHIs. Kluth et al. evaluated amorphous SiO2 by small angle X-ray scattering (SAXS) and reported the total track diameters of ~10 nm for low velocity and ~8 nm for high velocity, respectively28. It should be noted that tracks in amorphous SiO2 have core/shell structures15, which is different from the hard-cylinder-type of crystalline SiO230.

P. Kumar et al. have studied the Se dependence of track diameter in C60 films using TEM observation under various ions (Se = 0.04–40.8 keV/nm) irradiations. They observed the largest ion track of ~20 nm in diameter under 30 MeV C60 irradiation. Furthermore, they observed that the diameter increased in proportional to Se, which was different from the conventional one that the square of the diameter is proportional to the Se. To explain the dependence, Kumar et al., proposed the Coulomb explosion model for the track formation in C60 films35. Kitayama et al. studied the core/shell ion tracks in amorphous SiN film by C60 ion irradiation. The total and core diameters did not depend on Se and were almost constant of 11.2 nm and 4.3 nm between Se = 3 and 22 keV/nm36. Relatively large but unusual energy dependent track formations were reported, which could be related to the synergy effect between Se and Sn.

Possible applications

Another concern, i.e., rather pragmatic motivation, of this study is possible substitutability of swift heavy ions (SHIs) by MeV C60 ions. Acceleration of SHIs requires the world-class big accelerator facilities. The number of the facilities available for materials science applications is quite limited, e.g., a few facilities in Japan. Since a few to several MeV C60 ions provide high electronic energy deposition comparable to SHIs, they could be used as substitution of SHIs. The acceleration energy of a few MeV is not practically high. It is easily attainable in quite common accelerator facilities with the terminal voltage of ~1–2 MV, which are found at many places. If the newly-developed high flux C60 ion sources can be combined with the commonly-used ~1 MV accelerator facilities, the high-density electronic excitation which are comparable to SHIs would be easily accessible in quite many facilities, i.e., more than 20 facilities in Japan.

However, some irradiation effects of MeV C60 ions are different from those of SHIs. Ion ranges of SHIs are e.g., ~10 μm, much longer than those of MeV C60 ions, even assuming the ion range of C60 ion as that of constitute C monomer ion, e.g., ~a few hundred nm. As shown in this paper, the depths where the high Se is available are further limited, by the fragmentation of C60 ions. The cluster enhanced sputtering cannot be neglected. The perfect substitution of SHIs by MeV C60 ions is difficult. However, substitution of certain aspects could be possible. To understand the substitutability, we are studying the interaction between MeV C60 ions and materials, in this case, NPs.

Summary

Fullerene ions of 1–6 MeV induce dense electronic excitation in SiO2 comparable to swift heavy ions of 200 MeV Xe14+. Shape elongation of Au and Zn NPs was confirmed in amorphous SiO2 under the C60+ ion irradiation. According to the optical dichroism spectroscopy, 4 MeV and 6 MeV C60 ion irradiation induced Zn NP shape elongation with a high efficiency at low fluences, which was comparable to or slightly higher than 200 MeV Xe ions. While the Se of 6 MeV C60 ions and that of 200 MeV Xe ions were virtually the same, the slower velocity of the former reduced the δ-ray energy, which resulted in the enhancement of the excitation density26. While the NP elongation linearly increased with the fluence up to 1 × 1013 Xe14+/cm2 for 200 MeV Xe ions, the elongation by C60 ions sublinearly increased exceeding 1 × 1012 C60+/cm2 and decreased above 1 × 1013 C60+/cm2 due to the highly efficient sputtering. It should be noted that the elongation is induced even with 1 MeV C60 ions.

XTEM observations clearly showed not only the elongation of NPs but also a significant loss of the SiO2 layer. The significant loss was also confirmed by RBS and even the naked eye as loss of the NP layer’s color. Since Se increased and Sn decreased in the energy region between 1 and 6 MeV, the observed sputtering that increased with increasing energy was ascribed to the electronic origin. However, the observed magnitude was higher by more than one order than the empirical cubic law27, Y = B1Se3 for electronic sputtering by monomer SHIs, indicating the cluster enhancement and/or the synergy effect17,33,34.

Ion tracks and possible hillocks caused by C60 ion irradiation were observed in crystalline SiO2 by XTEM observations. The tracks of 10.4 ± 3.0 nm in diameter and 60–80 nm in length were formed under 4 MeV C60 ion irradiation. The track length was much shorter than those associated with SHIs having the same Se. To understand whether the elongation of NPs was induced in a layer deeper than the ion tracks, a SiO2 layer of 200 nm was deposited on the NP layer and irradiated with 6 MeV C60 ions. The NPs were rarely elongated in this configuration where the NPs were not touched by any ion tracks but impacted with C60 ions at ~2.5 MeV if they could not be fragmented. This observation confirmed that most of the C60 ions were fragmented beyond the track length. Since the fragments were too small and too separated in the region deeper than the track lengths, cooperative excitation between the fragments were no longer expected.

Methods

Samples

Two types of NP samples were prepared: (i) Au NPs were formed in amorphous SiO2 by sequential vacuum depositions and annealing: At first, 3 nm thick of Au film was deposited on SiO2 substrate by electron beam vaporization. Rapid thermal annealing at 300 °C for 10 minutes transformed the continuous Au film to isolated NPs. Then the NPs were covered with SiO2 film of 100 nm thick by sputtering deposition. See supplementary materials. (ii) Zn NPs were formed by implantation of 60 keV Zn ions to SiO2 to a fluence of 1 × 1017 Zn+/cm2. Even without post thermal annealing, Zn NPs of 10.3 ± 2.3 nm in diameter were formed in the depth region between 20 and 70 nm.

Irradiations

The irradiation of C60+ ions was carried out at Takasaki Advanced Radiation Research Institute (TARRI), National Institutes for Quantum and Radiological Science and Technology (QST), using a 3 MV tandem accelerator and a newly developed high-flux C60 negative ion source. While the samples were irradiated with C60 positive ion beams of four different energies, 1, 2, 4, and 6 MeV, the beam current was utilized at ~50 pA for 6 MeV and ~100 pA for other energies through an aperture of 3 mm in diameter. Some samples were irradiated with an incident angle of 45°, to detect the shape elongation of NPs by the optical linear dichroism (OLD) spectroscopy. Crystalline SiO2 samples were irradiated with 0 or 7° from the surface normal for ion-track observation by cross-sectional transmission electron microscopy (XTEM).

Measurements

A standard dual-beam spectrophotometer was used for the OLD spectroscopy in the wavelength region of 215–800 nm with a resolution of 1 nm; a pair of optical polarizers (extinction ratio < 5 × 10−5 each) were used. A sample was set between the two polarizers, P and A, and illuminated by linearly polarized monochromatic light from the spectrophotometer through the first polarizer P. Light transmitted through the sample was detected through the second polarizer A, whose polarization plane was set to the same angle as that of polarizer P. The role of the second polarizer is to remove the birefringence signal. The optical transmittance was plotted in the form of optical density (OD = −log10 T) without correction of reflection, where T denotes the transmittance. An area of approximately 1 mm in diameter of the sample was illuminated through an aperture.

XTEM was carried out using JEM-2100 (for bright field observation) and JEM-2100F, JEOL (for scanning TEM and X-ray energy dispersive spectrometry mapping) under an acceleration voltage of 200 kV. The XTEM samples were fabricated with 30 keV Ga focused ion beam (FIB) milling. To identify the surface position in the cross-sectional configuration, thin layer of Pt was deposited on the sample surface before the FIB milling. Rutherford backscattering spectrometry (RBS) was carried out in TARRI to determine the content of Zn atoms in the sample using a 2 MeV He+ beam of 1 mm in diameter with a scattering angle of 165°. The data were analyzed with RUMP code.